Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Re: Migrating from Evolution to Thunderbird
15 views
Skip to first unread message
Message has been deleted
Giovanna Stefani
Oct 20, 2011, 7:28:15 AM10/20/11
to
On Thu, 20 Oct 2011 10:50:46 +0000, Huge wrote:
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to
> migrate my email from Evolution to Thunderbird. (Anyone know if Mint is
> going to stop with X/Gnome?)
You don't say why you 'have to' migrate.
It is not something I would do willingly as I prefer Evolution over
Thunderbird any day.
Ubuntu is, after all, a mangled version of Debian. Have you tried Debian
6? It is a fine system and much better/easier than ubuntu.
Evolution backup and restore should cover your needs when changing OS.
--
Giovanna
anamedas.org.uk
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to
> migrate my email from Evolution to Thunderbird. (Anyone know if Mint is
> going to stop with X/Gnome?)
You don't say why you 'have to' migrate.
It is not something I would do willingly as I prefer Evolution over
Thunderbird any day.
Ubuntu is, after all, a mangled version of Debian. Have you tried Debian
6? It is a fine system and much better/easier than ubuntu.
Evolution backup and restore should cover your needs when changing OS.
--
Giovanna
anamedas.org.uk
Message has been deleted
Giovanna Stefani
Oct 20, 2011, 7:39:27 AM10/20/11
to
On Thu, 20 Oct 2011 11:31:16 +0000, Huge wrote:
> On 2011-10-20, Giovanna Stefani <giov...@notmail.com> wrote:
>> On Thu, 20 Oct 2011 10:50:46 +0000, Huge wrote:
>>
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to
>>> migrate my email from Evolution to Thunderbird. (Anyone know if Mint
>>> is going to stop with X/Gnome?)
>>
>> You don't say why you 'have to' migrate.
>
> Because I want my system to be as standard as possible. After 40 years
> of tinkering with computers it is not the kind of thing I want to do
> with my "day-to-day" machine. It's just a tool.
Yet you spent 'months' migrating from Thunderbird to Evolution. I wonder
why you think Thunderbird is somehow more 'standard' than Evolution when
the latter is pretty much the default on Gnome.
If it is just a tool, why keep tinkering (:
--
Giovanna
> On 2011-10-20, Giovanna Stefani <giov...@notmail.com> wrote:
>> On Thu, 20 Oct 2011 10:50:46 +0000, Huge wrote:
>>
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to
>>> migrate my email from Evolution to Thunderbird. (Anyone know if Mint
>>> is going to stop with X/Gnome?)
>>
>> You don't say why you 'have to' migrate.
>
> of tinkering with computers it is not the kind of thing I want to do
> with my "day-to-day" machine. It's just a tool.
Yet you spent 'months' migrating from Thunderbird to Evolution. I wonder
why you think Thunderbird is somehow more 'standard' than Evolution when
the latter is pretty much the default on Gnome.
If it is just a tool, why keep tinkering (:
--
Giovanna
A. Filip
Oct 20, 2011, 8:16:51 AM10/20/11
to
Huge <Hu...@nowhere.much.invalid> wrote:
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to migrate
> my email from Evolution to Thunderbird. (Anyone know if Mint is going
> to stop with X/Gnome?)
> [...]
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to migrate
> my email from Evolution to Thunderbird. (Anyone know if Mint is going
> to stop with X/Gnome?)
Have you considered migrating to local IMAP server?
[ e.g. to dovecot with mail stored in ~/Maildir ]
I use it myself because I do dislike being tied to single email client.
I also use fetchmail for fetching email from remote accounts and sieve
for sorting incoming mail.
I suspect migrating from an internal evolution email storage to IMAP
storage could be as easy as "drag and drop".
--
[pl>en Andrew] Andrzej A. Filip : an...@onet.eu : Andrze...@gmail.com
A penny saved is a penny to squander.
-- Ambrose Bierce
Message has been deleted
A. Filip
Oct 20, 2011, 9:37:11 AM10/20/11
to
Huge <Hu...@nowhere.much.invalid> wrote:
> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
> I should think about going back that way with something a little less
> consumptive ... ?
I run my "personal IMAP server/software" on 127.0.0.1 - It reduces fears
of additional electricity consumption/costs, does not it? ;-)
I use "loopback IMAP" mainly to be able to use a few different email
clients. I use emacs+gnus for mailing lists (as for usenet too) and
thunderbird/icedove for rest of my email. I had also used/tested for
some time kmail and evolution.
> (What's the emoticon for gazing into space, pondering?)
Certainly there are things in life that money can't buy,
But it's very funny -- did you ever try buying them without money?
-- Ogden Nash
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to migrate
>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>> to stop with X/Gnome?)
>>> [...]
>>
>> Have you considered migrating to local IMAP server?
>
> Interesting and constructive point, thank you.
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to migrate
>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>> to stop with X/Gnome?)
>>> [...]
>>
>> Have you considered migrating to local IMAP server?
>
>
>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>
>> I use it myself because I do dislike being tied to single email client.
>> I also use fetchmail for fetching email from remote accounts and sieve
>> for sorting incoming mail.
>>
>> I suspect migrating from an internal evolution email storage to IMAP
>> storage could be as easy as "drag and drop".
>
> I used to run a "house server" which did mail, NAS, etc., but stopped
>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>
>> I use it myself because I do dislike being tied to single email client.
>> I also use fetchmail for fetching email from remote accounts and sieve
>> for sorting incoming mail.
>>
>> I suspect migrating from an internal evolution email storage to IMAP
>> storage could be as easy as "drag and drop".
>
> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
> I should think about going back that way with something a little less
> consumptive ... ?
I run my "personal IMAP server/software" on 127.0.0.1 - It reduces fears
of additional electricity consumption/costs, does not it? ;-)
I use "loopback IMAP" mainly to be able to use a few different email
clients. I use emacs+gnus for mailing lists (as for usenet too) and
thunderbird/icedove for rest of my email. I had also used/tested for
some time kmail and evolution.
> (What's the emoticon for gazing into space, pondering?)
Certainly there are things in life that money can't buy,
But it's very funny -- did you ever try buying them without money?
-- Ogden Nash
Message has been deleted
The Nomad
Oct 20, 2011, 9:49:27 AM10/20/11
to
On Thu, 20 Oct 2011 13:42:30 +0000, Huge wrote:
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>>>> Huge <Hu...@nowhere.much.invalid> wrote:
>>>>> So, even if I stay with Ubuntu (unlikely, unless they go back on
>>>>> their decision to dump X and use Unity as the UI), I'm going to have
>>>>> to migrate my email from Evolution to Thunderbird. (Anyone know if
>>>>> Mint is going to stop with X/Gnome?)
>>>>> [...]
>>>>
>>>> Have you considered migrating to local IMAP server?
>>>
>>> Interesting and constructive point, thank you.
>>>
>>>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>>>
>>>> I use it myself because I do dislike being tied to single email
>>>> client. I also use fetchmail for fetching email from remote accounts
>>>> and sieve for sorting incoming mail.
>>>>
>>>> I suspect migrating from an internal evolution email storage to IMAP
>>>> storage could be as easy as "drag and drop".
>>>
>>> I used to run a "house server" which did mail, NAS, etc., but stopped
>>> because of the electricity consumption. (Mind you, it was a Sun).
>>> Perhaps I should think about going back that way with something a
>>> little less consumptive ... ?
>>
>> I run my "personal IMAP server/software" on 127.0.0.1 - It reduces
>> fears of additional electricity consumption/costs, does not it? ;-)
>
> Unfortunately, I have multiple users and multiple machines here. It adds
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>>>> Huge <Hu...@nowhere.much.invalid> wrote:
>>>>> So, even if I stay with Ubuntu (unlikely, unless they go back on
>>>>> their decision to dump X and use Unity as the UI), I'm going to have
>>>>> to migrate my email from Evolution to Thunderbird. (Anyone know if
>>>>> Mint is going to stop with X/Gnome?)
>>>>> [...]
>>>>
>>>> Have you considered migrating to local IMAP server?
>>>
>>> Interesting and constructive point, thank you.
>>>
>>>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>>>
>>>> I use it myself because I do dislike being tied to single email
>>>> client. I also use fetchmail for fetching email from remote accounts
>>>> and sieve for sorting incoming mail.
>>>>
>>>> I suspect migrating from an internal evolution email storage to IMAP
>>>> storage could be as easy as "drag and drop".
>>>
>>> I used to run a "house server" which did mail, NAS, etc., but stopped
>>> because of the electricity consumption. (Mind you, it was a Sun).
>>> Perhaps I should think about going back that way with something a
>>> little less consumptive ... ?
>>
>> I run my "personal IMAP server/software" on 127.0.0.1 - It reduces
>> fears of additional electricity consumption/costs, does not it? ;-)
>
> complexity and complication (and electricity!) if I say to my wife that
> she has to switch on my computer as well as hers in order to read her
> mail! It would certainly be helpful to be able to read my mail on my
> netbook. I'm beginning to warm to this idea.
IANAE!
I run an atom based machine (in the loft) doing just that (dovecot/
fetchmail) and nas stuff as well.
Not sure of the consumption but <60W I guess.
N
Message has been deleted
A. Filip
Oct 20, 2011, 10:52:58 AM10/20/11
to
Huge <Hu...@nowhere.much.invalid> wrote:
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>>>> Huge <Hu...@nowhere.much.invalid> wrote:
>>>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>>>> decision to dump X and use Unity as the UI), I'm going to have to migrate
>>>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>>>> to stop with X/Gnome?)
>>>>> [...]
>>>>
>>>> Have you considered migrating to local IMAP server?
>>>
>>> Interesting and constructive point, thank you.
>>>
>>>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>>>
>>>> I use it myself because I do dislike being tied to single email client.
>>>> I also use fetchmail for fetching email from remote accounts and sieve
>>>> for sorting incoming mail.
>>>>
>>>> I suspect migrating from an internal evolution email storage to IMAP
>>>> storage could be as easy as "drag and drop".
>>>
>>> I used to run a "house server" which did mail, NAS, etc., but stopped
>>> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
>>> I should think about going back that way with something a little less
>>> consumptive ... ?
>>
>> I run my "personal IMAP server/software" on 127.0.0.1 - It reduces fears
>> of additional electricity consumption/costs, does not it? ;-)
>
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>>>> Huge <Hu...@nowhere.much.invalid> wrote:
>>>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>>>> decision to dump X and use Unity as the UI), I'm going to have to migrate
>>>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>>>> to stop with X/Gnome?)
>>>>> [...]
>>>>
>>>> Have you considered migrating to local IMAP server?
>>>
>>> Interesting and constructive point, thank you.
>>>
>>>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>>>
>>>> I use it myself because I do dislike being tied to single email client.
>>>> I also use fetchmail for fetching email from remote accounts and sieve
>>>> for sorting incoming mail.
>>>>
>>>> I suspect migrating from an internal evolution email storage to IMAP
>>>> storage could be as easy as "drag and drop".
>>>
>>> I used to run a "house server" which did mail, NAS, etc., but stopped
>>> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
>>> I should think about going back that way with something a little less
>>> consumptive ... ?
>>
>> I run my "personal IMAP server/software" on 127.0.0.1 - It reduces fears
>> of additional electricity consumption/costs, does not it? ;-)
>
> Unfortunately, I have multiple users and multiple machines here. It adds
> complexity and complication (and electricity!) if I say to my wife that she
> has to switch on my computer as well as hers in order to read her mail! It
> would certainly be helpful to be able to read my mail on my netbook. I'm
> beginning to warm to this idea.
How do you intend to get email from external accounts?
> complexity and complication (and electricity!) if I say to my wife that she
> has to switch on my computer as well as hers in order to read her mail! It
> would certainly be helpful to be able to read my mail on my netbook. I'm
> beginning to warm to this idea.
I use periodic fetchmail polling but
* you may consider using IMAP IDLE for getting messages with negligible
delays. AFAIK gmail supports IMAP IDLE but I am not quite sure if
24h/day IMAP sessions are going to be tolerated :-)
* in case of periodic polling some way to allow user force "extra polling"
would be handy IMHO [e.g. via custom command executed via ssh session]
You just wait, I'll sin till I blow up!
-- Dylan Thomas
The Nomad
Oct 20, 2011, 1:29:42 PM10/20/11
to
On Thu, 20 Oct 2011 14:50:51 +0000, Huge wrote:
> I am not an ... ?
unknown quantity + drip under pressure :-) (3.142-ish)
> And since my central heating is broken at the moment, I could do with
> the heat...
Snap!
(3.142-ish) Expert.
unknown quantity + drip under pressure :-) (3.142-ish)
> And since my central heating is broken at the moment, I could do with
> the heat...
Snap!
(3.142-ish) Expert.
Nix
Oct 20, 2011, 1:39:55 PM10/20/11
to
On 20 Oct 2011, Hu...@nowhere.much.invalid told this:
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to migrate
> my email from Evolution to Thunderbird. (Anyone know if Mint is going
> to stop with X/Gnome?)
They're not dumping X. Unity is just an alternative wm and replacement
for the GNOME panel, that's all.
--
NULL && (void)
> So, even if I stay with Ubuntu (unlikely, unless they go back on their
> decision to dump X and use Unity as the UI), I'm going to have to migrate
> my email from Evolution to Thunderbird. (Anyone know if Mint is going
> to stop with X/Gnome?)
for the GNOME panel, that's all.
--
NULL && (void)
Nix
Oct 20, 2011, 1:42:17 PM10/20/11
to
On 20 Oct 2011, The Nomad uttered the following:
Get a Soekris net5501 or something. <3W if you go for Flash, perhaps 8W
if you go for spinning rust.
(I use one, but only for firewalling / mailservice / bastion hosting. I
have a great big quad-core always-on headless RAID-array monster box for
most of my work, but that's been optimized for power consumption too to
some degree and draws a lot less than a random old two-disk Athlon ever
used to.)
--
NULL && (void)
> I run an atom based machine (in the loft) doing just that (dovecot/
> fetchmail) and nas stuff as well.
>
> Not sure of the consumption but <60W I guess.
Christ that's a lot.
> fetchmail) and nas stuff as well.
>
> Not sure of the consumption but <60W I guess.
Get a Soekris net5501 or something. <3W if you go for Flash, perhaps 8W
if you go for spinning rust.
(I use one, but only for firewalling / mailservice / bastion hosting. I
have a great big quad-core always-on headless RAID-array monster box for
most of my work, but that's been optimized for power consumption too to
some degree and draws a lot less than a random old two-disk Athlon ever
used to.)
--
NULL && (void)
Message has been deleted
Message has been deleted
Message has been deleted
Johny B Good
Oct 20, 2011, 7:49:26 PM10/20/11
to
On Thu, 20 Oct 2011 14:23:30 +0100, Huge <Hu...@nowhere.much.invalid> wrote:
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
> On 2011-10-20, A. Filip <an...@onet.eu> wrote:
>> Huge <Hu...@nowhere.much.invalid> wrote:
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to
>>> migrate
>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>> to stop with X/Gnome?)
>>> [...]
>>
>> Have you considered migrating to local IMAP server?
>
>>> So, even if I stay with Ubuntu (unlikely, unless they go back on their
>>> decision to dump X and use Unity as the UI), I'm going to have to
>>> migrate
>>> my email from Evolution to Thunderbird. (Anyone know if Mint is going
>>> to stop with X/Gnome?)
>>> [...]
>>
>> Have you considered migrating to local IMAP server?
>
> Interesting and constructive point, thank you.
>
>
>> [ e.g. to dovecot with mail stored in ~/Maildir ]
>>
>> I use it myself because I do dislike being tied to single email client.
>> I also use fetchmail for fetching email from remote accounts and sieve
>> for sorting incoming mail.
>>
>> I suspect migrating from an internal evolution email storage to IMAP
>> storage could be as easy as "drag and drop".
>
>>
>> I use it myself because I do dislike being tied to single email client.
>> I also use fetchmail for fetching email from remote accounts and sieve
>> for sorting incoming mail.
>>
>> I suspect migrating from an internal evolution email storage to IMAP
>> storage could be as easy as "drag and drop".
>
> I used to run a "house server" which did mail, NAS, etc., but stopped
> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
> I should think about going back that way with something a little less
> consumptive ... ?
This might help you decide:
> because of the electricity consumption. (Mind you, it was a Sun). Perhaps
> I should think about going back that way with something a little less
> consumptive ... ?
I'm using an ancient Gateway 2000 desktop case with its original 145W ATX
PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This takes
between 46 and 52 watts when idle (no spin down power saving involved)
depending on what 'mood' the two 'so called eco green' Western Digital
WD20EARS drives[2] happen to be in.
The MoBo is nothing special, just a cheap Asrock microATX AM2 board with
a GB lan port and 4 SATA2 ports with a Sempron LE-1250 processor and 2GB
DDR2 RAM fitted. When I first set the MoBo up, I chose to forego the Cool
'n' Quiet option and use a fixed underclock and undervolt but I've since
discovered that FreeNAS does a better job of power management with the
Cool 'n' Quiet option enabled.
>
> (What's the emoticon for gazing into space, pondering?)
[1] You might think that an old small form factor 145W rated ATX PSU might
succumb to overload on its 4.2A rated 12v rail with all those hard disk
drives but it takes the spin up loading in its stride (even with the
additional load from the add-on 4 pin 12v connector lead for the CPU VRM).
I guess the 145W rating is the safe maximum CONTINUOUS rating rather than
a peak rating that would guarantee disaster if applied for more than 8
seconds on a typical 450 watts or greater rated ATX unit.
I stuck with the original PSU on account of its better efficiency and
thermally regulated fan which provided all of the (very quiet) cooling
required of the whole box. In fact I was so determined to retain it, I
even added the internals of a wallwart 5.2v smpsu to beef up the 100mA
rated 5VSB to a level that would satisfy the needs of a standard MoBo.
However, I do have a suitably miniature 270W ATX PSU in my "Server Spares"
box in the event of PSU failure (it's staying there until needed on
account it would add a watt or three to the server's consumption - not as
bad as a standard ATX PSU which could add as much as an extra ten watts).
[2] I've used the WDIDLE3 utility to increase the head unloading timeout
from the egregiously short 8.3 seconds to a more reasonable 5 minutes (you
can't actually totally disable this feature on these drives). I prefer not
to use spindown timeout power saving on hard disk drives, both for the
sake of my sanity and for the integrity of my data, otherwise I could
reduce idle consumption down to a mere 18 watts or so if I were to be so
cavalier.
The other two drives are Samsung HD203WI eco green drives which seem to
function 'as described'. I know that the weird power consumption curve is
down to those Western Digitals and not the Samsungs from observation of
the temperature data.
The Western Digitals were my second choice in the 2TB eco green drive
short list and were acquired to satisfy a critical capacity upgrade
otherwise I'd have only used Samsungs. Now, of course, Samsung is no
longer an option in the 3TB and larger eco green drive capacities that I'm
now having to consider for the next capacity upgrade. I'm hoping against
hope that the next Western Digital eco green drives will be much better
behaved than those WD20EARS units.
--
Regards JB Good
Mike Scott
Oct 21, 2011, 3:44:55 AM10/21/11
to
On 20/10/2011 15:52, A. Filip wrote:
....
mail server, and set up the MX records for your domain accordingly.
Works quite well enough for a domestic system on a dynamic ip.
Side effect - you can lock out an awful lot of spammers with your own
firewall.
And switch ISP without having to change your email address.
(I use sendmail/dovecot running under freebsd on a C7-based box with a
2.5" drive. Not speedy, but quiet and fairly low power. It serves the 3
or 4 PCs that we use at home.)
--
Mike Scott (unet2 <at> [deletethis] scottsonline.org.uk)
Harlow Essex England
....
>> Unfortunately, I have multiple users and multiple machines here. It adds
>> complexity and complication (and electricity!) if I say to my wife that she
>> has to switch on my computer as well as hers in order to read her mail! It
>> would certainly be helpful to be able to read my mail on my netbook. I'm
>> beginning to warm to this idea.
>
> How do you intend to get email from external accounts?
>
> I use periodic fetchmail polling but
> * you may consider using IMAP IDLE for getting messages with negligible
> delays. AFAIK gmail supports IMAP IDLE but I am not quite sure if
> 24h/day IMAP sessions are going to be tolerated :-)
> * in case of periodic polling some way to allow user force "extra polling"
> would be handy IMHO [e.g. via custom command executed via ssh session]
>
Or use your own domain name. Use dyndns so you have a handle to your
>> complexity and complication (and electricity!) if I say to my wife that she
>> has to switch on my computer as well as hers in order to read her mail! It
>> would certainly be helpful to be able to read my mail on my netbook. I'm
>> beginning to warm to this idea.
>
> How do you intend to get email from external accounts?
>
> I use periodic fetchmail polling but
> * you may consider using IMAP IDLE for getting messages with negligible
> delays. AFAIK gmail supports IMAP IDLE but I am not quite sure if
> 24h/day IMAP sessions are going to be tolerated :-)
> * in case of periodic polling some way to allow user force "extra polling"
> would be handy IMHO [e.g. via custom command executed via ssh session]
>
mail server, and set up the MX records for your domain accordingly.
Works quite well enough for a domestic system on a dynamic ip.
Side effect - you can lock out an awful lot of spammers with your own
firewall.
And switch ISP without having to change your email address.
(I use sendmail/dovecot running under freebsd on a C7-based box with a
2.5" drive. Not speedy, but quiet and fairly low power. It serves the 3
or 4 PCs that we use at home.)
--
Mike Scott (unet2 <at> [deletethis] scottsonline.org.uk)
Harlow Essex England
Gordon
Oct 21, 2011, 3:50:46 AM10/21/11
to
Running X alone is not at all useful, for there is nothing on the screen.
X is nothing more than some windows which the window manager usesto produce
some slightly more useful stuff.
Mint is a distro, Gnome is a desktop. They can some times be seen together,
but there is no need for them to do so. Mint also has other desktops which
work just fine for some people, and great for others.
You have a choice, some home work is required to get the best for one.
Enjoy
Gordon
Oct 21, 2011, 3:52:47 AM10/21/11
to
Giovanna Stefani
Oct 21, 2011, 6:05:17 AM10/21/11
to
On Fri, 21 Oct 2011 07:52:47 +0000, Gordon wrote:
> No Ms Penguin, the Troll bell has not gone off.
And a good thing too. Happily, there seem to be few, if any, on this
> No Ms Penguin, the Troll bell has not gone off.
group.
--
Giovanna
Debian 6 user
Aragorn
Oct 21, 2011, 6:44:19 AM10/21/11
to
On Friday 21 October 2011 12:05, Giovanna Stefani conveyed the following
to uk.comp.os.linux...
> On Fri, 21 Oct 2011 07:52:47 +0000, Gordon wrote:
>
>> No Ms Penguin, the Troll bell has not gone off.
>
> And a good thing too. Happily, there seem to be few, if any, on this
> group.
Yeah, they're all camping in alt.os.linux.ubuntu, and on occasion the
inmates of comp.os.linux.advocacy are allowed to temporarily leave their
loony bin and then post their same old MickeySoft fanboi nonsense to
some of the comp.os.linux.* groups, or to alt.os.linux. They've tried
it here already too, though.
Oh well, killfiles are a great idea. ;-)
--
= Aragorn =
(registered GNU/Linux user #223157)
to uk.comp.os.linux...
> On Fri, 21 Oct 2011 07:52:47 +0000, Gordon wrote:
>
>> No Ms Penguin, the Troll bell has not gone off.
>
> And a good thing too. Happily, there seem to be few, if any, on this
> group.
inmates of comp.os.linux.advocacy are allowed to temporarily leave their
loony bin and then post their same old MickeySoft fanboi nonsense to
some of the comp.os.linux.* groups, or to alt.os.linux. They've tried
it here already too, though.
Oh well, killfiles are a great idea. ;-)
--
= Aragorn =
(registered GNU/Linux user #223157)
Gordon
Oct 21, 2011, 7:47:01 AM10/21/11
to
On 20/10/11 12:39, Giovanna Stefani wrote:
>
> Yet you spent 'months' migrating from Thunderbird to Evolution. I wonder
> why you think Thunderbird is somehow more 'standard' than Evolution when
> the latter is pretty much the default on Gnome.
> If it is just a tool, why keep tinkering (:
>
>
B4ecause in Ubuntu 11.10 Thunderbird IS the default email client - like
>
> Yet you spent 'months' migrating from Thunderbird to Evolution. I wonder
> why you think Thunderbird is somehow more 'standard' than Evolution when
> the latter is pretty much the default on Gnome.
> If it is just a tool, why keep tinkering (:
>
>
Evolution was before 11.10...
--
My Alternative Computing Blog <http://gbplinuxfoss.blogspot.com/>
Giovanna Stefani
Oct 21, 2011, 8:33:06 AM10/21/11
to
On Fri, 21 Oct 2011 12:47:01 +0100, Gordon wrote:
> On 20/10/11 12:39, Giovanna Stefani wrote:
>
>
>> Yet you spent 'months' migrating from Thunderbird to Evolution. I
>> wonder why you think Thunderbird is somehow more 'standard' than
>> Evolution when the latter is pretty much the default on Gnome. If it is
>> just a tool, why keep tinkering (:
>>
>>
>>
> Because in Ubuntu 11.10 Thunderbird IS the default email client - like
> On 20/10/11 12:39, Giovanna Stefani wrote:
>
>
>> Yet you spent 'months' migrating from Thunderbird to Evolution. I
>> wonder why you think Thunderbird is somehow more 'standard' than
>> Evolution when the latter is pretty much the default on Gnome. If it is
>> just a tool, why keep tinkering (:
>>
>>
>>
> Evolution was before 11.10...
I was (obviously) unaware of that, thank you.
One assumes that Evolution is still available as an option in which case
there would be no real need to change one's own preferences simply
because one distro has.
I did use Ubuntu briefly several years ago when I first migrated to Linux
from Windows but I soon became disillusioned with its attempts to 'be'
Windows (or so I perceived its development). After trying various distros,
I moved on to Debian 4 and have stayed with Debian through versions 5 and
(currently) 6.
Thanks again for clarifying the default issue.
Richard Kettlewell
Oct 21, 2011, 8:51:55 AM10/21/11
to
Giovanna Stefani <giov...@notmail.com> writes:
> Gordon wrote:
>> On 20/10/11 12:39, Giovanna Stefani wrote:
>>> Yet you spent 'months' migrating from Thunderbird to Evolution. I
>>> wonder why you think Thunderbird is somehow more 'standard' than
>>> Evolution when the latter is pretty much the default on Gnome. If it is
>>> just a tool, why keep tinkering (:
>>
>> Because in Ubuntu 11.10 Thunderbird IS the default email client - like
>> Evolution was before 11.10...
>
> I was (obviously) unaware of that, thank you.
> One assumes that Evolution is still available as an option in which case
> there would be no real need to change one's own preferences simply
> because one distro has.
Yes, the package still exists in oneiric. Judging by
> Gordon wrote:
>> On 20/10/11 12:39, Giovanna Stefani wrote:
>>> Yet you spent 'months' migrating from Thunderbird to Evolution. I
>>> wonder why you think Thunderbird is somehow more 'standard' than
>>> Evolution when the latter is pretty much the default on Gnome. If it is
>>> just a tool, why keep tinkering (:
>>
>> Because in Ubuntu 11.10 Thunderbird IS the default email client - like
>> Evolution was before 11.10...
>
> I was (obviously) unaware of that, thank you.
> One assumes that Evolution is still available as an option in which case
> there would be no real need to change one's own preferences simply
> because one distro has.
http://packages.ubuntu.com/oneiric/ubuntu-desktop it is not installed by
default however.
Some of the online traffic concerning the change does mention migration
but if there's anything more automated than the instructions on the
wiki[1] then I couldn't find them.
[1] https://help.ubuntu.com/community/MigrateEvolutionToThunderbird
--
http://www.greenend.org.uk/rjk/
Message has been deleted
Tony Houghton
Oct 21, 2011, 10:18:21 AM10/21/11
to
In <878vofp...@spindle.srvr.nix>,
Long-term they are planning to replace X with something called Wayland.
--
TH * http://www.realh.co.uk
--
TH * http://www.realh.co.uk
Nix
Oct 21, 2011, 10:33:15 AM10/21/11
to
On 21 Oct 2011, Johny B. Good outgrape:
drives in my RAID array, half the size of yours and still nowhere near
full). They don't spin all the way down, though, just to about 2000rpm,
then back up if I/O becomes frequent enough. They keep servicing
requests the whole time. Sodding awesome transfer rate with four bound
together over PCIe too: between 190MiB/s at high sectors up to 250MiB/s
at low. My previous aged PCI software RAID array could manage 30Mb/s if
you were very lucky...
(The machine itself is not ancient but a two-year-old headless Nehalem
server box, 24Gb RAM and all: I do anything that requires serious
gruntwork on it.)
> [1] You might think that an old small form factor 145W rated ATX PSU
> might succumb to overload on its 4.2A rated 12v rail with all those
> hard disk drives but it takes the spin up loading in its stride (even
Yeah, four disks are unlikely to cause problems. Eight might though. (My
RAID array is a hardware one and has an option to stagger spinup
specifically to avoid overloading the PSU.)
> I stuck with the original PSU on account of its better efficiency and
> thermally regulated fan which provided all of the (very quiet) cooling
> required of the whole box.
One would hope newer PSUs also have fans that don't spin at top speed
the whole time, given that they sound like jet engines when they do.
> [2] I've used the WDIDLE3 utility to increase the head unloading
> timeout from the egregiously short 8.3 seconds to a more reasonable 5
> minutes
Interesting. The heads don't unload on mine at all: this is even touted
as a feature ('cos it means you can always get at your data with zero
delay). Maybe yours are laptop drives or something?
> (you can't actually totally disable this feature on these
> drives). I prefer not to use spindown timeout power saving on hard
> disk drives,
I do it simply because it saves me power and thus money. It is also
reasonable to assume that the drives are expected to operate mostly at
the lower speed, so continuous high-speed spinning might shorten their
life.
> The Western Digitals were my second choice in the 2TB eco green drive
> short list and were acquired to satisfy a critical capacity upgrade
> otherwise I'd have only used Samsungs. Now, of course, Samsung is no
> longer an option in the 3TB and larger eco green drive capacities that
> I'm now having to consider for the next capacity upgrade. I'm hoping
> against hope that the next Western Digital eco green drives will be
> much better behaved than those WD20EARS units.
I see no behaviour problems with my WD10EACS. What misbehaviour do you
see? Does it actually stop sending you data and spin all the way down
after only 8.3s? That seems very unlikely: are you sure you didn't
misinterpret the spindown-to-2k-RPM setting as a complete spindown?
--
NULL && (void)
> I'm using an ancient Gateway 2000 desktop case with its original 145W
> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
> takes between 46 and 52 watts when idle (no spin down power saving
> involved) depending on what 'mood' the two 'so called eco green'
> Western Digital WD20EARS drives[2] happen to be in.
These drives do spin down to save power (I know, I have four WD10EACS
> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
> takes between 46 and 52 watts when idle (no spin down power saving
> involved) depending on what 'mood' the two 'so called eco green'
> Western Digital WD20EARS drives[2] happen to be in.
drives in my RAID array, half the size of yours and still nowhere near
full). They don't spin all the way down, though, just to about 2000rpm,
then back up if I/O becomes frequent enough. They keep servicing
requests the whole time. Sodding awesome transfer rate with four bound
together over PCIe too: between 190MiB/s at high sectors up to 250MiB/s
at low. My previous aged PCI software RAID array could manage 30Mb/s if
you were very lucky...
(The machine itself is not ancient but a two-year-old headless Nehalem
server box, 24Gb RAM and all: I do anything that requires serious
gruntwork on it.)
> [1] You might think that an old small form factor 145W rated ATX PSU
> might succumb to overload on its 4.2A rated 12v rail with all those
> hard disk drives but it takes the spin up loading in its stride (even
RAID array is a hardware one and has an option to stagger spinup
specifically to avoid overloading the PSU.)
> I stuck with the original PSU on account of its better efficiency and
> thermally regulated fan which provided all of the (very quiet) cooling
> required of the whole box.
the whole time, given that they sound like jet engines when they do.
> [2] I've used the WDIDLE3 utility to increase the head unloading
> timeout from the egregiously short 8.3 seconds to a more reasonable 5
> minutes
as a feature ('cos it means you can always get at your data with zero
delay). Maybe yours are laptop drives or something?
> (you can't actually totally disable this feature on these
> drives). I prefer not to use spindown timeout power saving on hard
> disk drives,
reasonable to assume that the drives are expected to operate mostly at
the lower speed, so continuous high-speed spinning might shorten their
life.
> The Western Digitals were my second choice in the 2TB eco green drive
> short list and were acquired to satisfy a critical capacity upgrade
> otherwise I'd have only used Samsungs. Now, of course, Samsung is no
> longer an option in the 3TB and larger eco green drive capacities that
> I'm now having to consider for the next capacity upgrade. I'm hoping
> against hope that the next Western Digital eco green drives will be
> much better behaved than those WD20EARS units.
see? Does it actually stop sending you data and spin all the way down
after only 8.3s? That seems very unlikely: are you sure you didn't
misinterpret the spindown-to-2k-RPM setting as a complete spindown?
--
NULL && (void)
Aragorn
Oct 21, 2011, 10:47:38 AM10/21/11
to
On Friday 21 October 2011 16:18, Tony Houghton conveyed the following to
uk.comp.os.linux...
... Which runs in kernelspace, and which is (as yet) not network-
transparent, unlike X11.
As for Unity, it's essentially a shell on top of Gnome 3.x, and it is
open source, but Canonical demands that anyone developing code for Unity
would sign over the copyrights to them.
It is however (at this stage) still possible to abandon Unity in Ubuntu
and go back to Gnome 2.3. Of course, this is only temporarily possible.
Gnome 2.x is no longer maintained, so in the long run, Ubuntu proper
will only offer Unity; there will of course also always be Kubuntu (with
KDE), Xubuntu (with XFCE) and Lubuntu (with LXDE).
uk.comp.os.linux...
transparent, unlike X11.
As for Unity, it's essentially a shell on top of Gnome 3.x, and it is
open source, but Canonical demands that anyone developing code for Unity
would sign over the copyrights to them.
It is however (at this stage) still possible to abandon Unity in Ubuntu
and go back to Gnome 2.3. Of course, this is only temporarily possible.
Gnome 2.x is no longer maintained, so in the long run, Ubuntu proper
will only offer Unity; there will of course also always be Kubuntu (with
KDE), Xubuntu (with XFCE) and Lubuntu (with LXDE).
Tony Mountifield
Oct 21, 2011, 1:16:59 PM10/21/11
to
In article <j7s0mb$j2c$2...@dont-email.me>,
that the Windows graphics subsystem ran in kernel space, which was one
of the reasons it could be unstable...
Cheers
Tony
--
Tony Mountifield
Work: to...@softins.co.uk - http://www.softins.co.uk
Play: to...@mountifield.org - http://tony.mountifield.org
Aragorn <str...@telenet.be.invalid> wrote:
> On Friday 21 October 2011 16:18, Tony Houghton conveyed the following to
> uk.comp.os.linux...
>
> On Friday 21 October 2011 16:18, Tony Houghton conveyed the following to
> uk.comp.os.linux...
>
> > Long-term they are planning to replace X with something called
> > Wayland.
>
> ... Which runs in kernelspace, and which is (as yet) not network-
> transparent, unlike X11.
I remember once upon a time, one of the sticks used for MS-bashing was
> > Wayland.
>
> ... Which runs in kernelspace, and which is (as yet) not network-
> transparent, unlike X11.
that the Windows graphics subsystem ran in kernel space, which was one
of the reasons it could be unstable...
Cheers
Tony
--
Tony Mountifield
Work: to...@softins.co.uk - http://www.softins.co.uk
Play: to...@mountifield.org - http://tony.mountifield.org
Justin C
Oct 21, 2011, 2:20:05 PM10/21/11
to
On 2011-10-21, Mike Scott <usen...@scottsonline.org.uk.invalid> wrote:
[snip]
and Exim handles most of the spam rejection, if I can reject it at the
firewall then I'd like to do that.
Justin.
[snip]
>
> Or use your own domain name. Use dyndns so you have a handle to your
> mail server, and set up the MX records for your domain accordingly.
> Works quite well enough for a domestic system on a dynamic ip.
>
> Side effect - you can lock out an awful lot of spammers with your own
> firewall.
I'd be interested to know how you do this. I host mail for my own domain
> Or use your own domain name. Use dyndns so you have a handle to your
> mail server, and set up the MX records for your domain accordingly.
> Works quite well enough for a domestic system on a dynamic ip.
>
> Side effect - you can lock out an awful lot of spammers with your own
> firewall.
and Exim handles most of the spam rejection, if I can reject it at the
firewall then I'd like to do that.
Justin.
A. Filip
Oct 21, 2011, 3:21:29 PM10/21/11
to
a) block connections from nets with highest zombies infection rate
b) reduce system load by implementing "cheap" incoming SMTP connections
rate limits (may work well with grey-listing)
Spamhaus provides lists of nets "recommended for firewalling out"
http://www.spamhaus.org/drop/
The most disagreeable thing that your worst enemy says to your face does
not approach what your best friends say behind your back.
-- Alfred De Musset
Chris Davies
Oct 21, 2011, 4:39:46 PM10/21/11
to
Justin C <justi...@purestblue.com> wrote:
> I'd be interested to know how you do this. I host mail for my own domain
> and Exim handles most of the spam rejection, if I can reject it at the
> firewall then I'd like to do that.
What I find works for me is this combination -
> I'd be interested to know how you do this. I host mail for my own domain
> and Exim handles most of the spam rejection, if I can reject it at the
> firewall then I'd like to do that.
1. Fail2ban. (This is the firewall bit.) I have a rule that picks up 3
failed delivery attempts by the same host sender and unilaterally blocks
that host for 36 hours. This has become much less effective over the
last six months: I'd suggest it's probably as low as 1% effective.
2. Tweaks to exim. I like the "don't talk to me until I've said hello"
option, as it still gets a fair number of hits from impatient spammers.
3. Greylisting. If the sender actually has a valid recipient, but
they've not sent email to that recipient previously I'll reject for a few
minutes. Some people reckon that this is not so useful these days, but
statistically it still appears to knock out a significant number of
attempts.
4. Spamassassin and Antivirus. If you get past everything else, I've got
some custom rules in SA that mark up particular specific traits. I'm
also not particularly keen on senders without (decent) rDNS, and tend
to mark them up significantly.
I've a low volume system for me and my family. Over the last three months
I would suggest that greylisting is still the most effective deterrent by
volume per use of resource. Over one particular period I had 33990 SMTP
attempts. 90 hosts were banned by the firewall (so that's only around
3*90 = 270 of those attempts). 10200 of the attempts were initially
greylisted, with only 1020 of those being reoffered and accepted. Of
the 3780 messages in the period that passed some form of greylisting,
360 of them were completely rejected by SpamAssassin or the AV.
I don't track repeated hits on the firewall after a host has been
banned. That might be interesting to measure.
Chris
alexd
Oct 21, 2011, 5:05:18 PM10/21/11
to
--
<http://ale.cx/> (AIM:troffasky) (UnSoEs...@ale.cx)
22:04:26 up 35 days, 3:17, 5 users, load average: 0.08, 0.24, 0.36
"People believe any quote they read on the internet
if it fits their preconceived notions." - Martin Luther King
alexd
Oct 21, 2011, 5:11:49 PM10/21/11
to
Tony Mountifield (for it is he) wrote:
> I remember once upon a time, one of the sticks used for MS-bashing was
> that the Windows graphics subsystem ran in kernel space, which was one
> of the reasons it could be unstable...
An unbeliever! Persecute! Kill!
> I remember once upon a time, one of the sticks used for MS-bashing was
> that the Windows graphics subsystem ran in kernel space, which was one
> of the reasons it could be unstable...
22:11:28 up 35 days, 3:24, 5 users, load average: 0.10, 0.16, 0.29
Nick Leverton
Oct 21, 2011, 5:46:43 PM10/21/11
to
In article <1381015.U...@ale.cx>, alexd <trof...@hotmail.com> wrote:
>Tony Mountifield (for it is he) wrote:
>
>> I remember once upon a time, one of the sticks used for MS-bashing was
>> that the Windows graphics subsystem ran in kernel space, which was one
>> of the reasons it could be unstable...
>
>An unbeliever! Persecute! Kill!
I'd been quite impressed with the idea of Wayland, but I didn't realise
>Tony Mountifield (for it is he) wrote:
>
>> I remember once upon a time, one of the sticks used for MS-bashing was
>> that the Windows graphics subsystem ran in kernel space, which was one
>> of the reasons it could be unstable...
>
>An unbeliever! Persecute! Kill!
it runs in kernel space. As a user task it would be quite a good idea.
Nick
--
Serendipity: http://www.leverton.org/blosxom (last update 29th March 2010)
"The Internet, a sort of ersatz counterfeit of real life"
-- Janet Street-Porter, BBC2, 19th March 1996
Martin Gregorie
Oct 21, 2011, 7:08:15 PM10/21/11
to
On Thu, 20 Oct 2011 16:52:58 +0200, A. Filip wrote:
> I use periodic fetchmail polling but * you may consider using IMAP IDLE
> for getting messages with negligible
>
I used to use it, then got pissed off by the way it left read-but-not-
> I use periodic fetchmail polling but * you may consider using IMAP IDLE
> for getting messages with negligible
>
deleted mail in my ISP mailbox forever or I manually deleted it. Since
then I switched to using getmail, which 'just works' and even uses the
same mda chain. Recommended. my distro, Fedora, includes it as a standard
package. Yours may too.
--
martin@ | Martin Gregorie
gregorie. | Essex, UK
org |
Martin Gregorie
Oct 21, 2011, 7:18:25 PM10/21/11
to
On Fri, 21 Oct 2011 21:39:46 +0100, Chris Davies wrote:
> Justin C <justi...@purestblue.com> wrote:
>> I'd be interested to know how you do this. I host mail for my own
>> domain and Exim handles most of the spam rejection, if I can reject it
>> at the firewall then I'd like to do that.
>
> What I find works for me is this combination -
>
> 1. Fail2ban. (This is the firewall bit.) I have a rule that picks up 3
> failed delivery attempts by the same host sender and unilaterally blocks
> that host for 36 hours. This has become much less effective over the
> last six months: I'd suggest it's probably as low as 1% effective.
>
By using getmail to collect incoming email from a POP3 mailbox at my ISB
> Justin C <justi...@purestblue.com> wrote:
>> I'd be interested to know how you do this. I host mail for my own
>> domain and Exim handles most of the spam rejection, if I can reject it
>> at the firewall then I'd like to do that.
>
> What I find works for me is this combination -
>
> 1. Fail2ban. (This is the firewall bit.) I have a rule that picks up 3
> failed delivery attempts by the same host sender and unilaterally blocks
> that host for 36 hours. This has become much less effective over the
> last six months: I'd suggest it's probably as low as 1% effective.
>
I don't need any holes in my firewall.
> 2. Tweaks to exim. I like the "don't talk to me until I've said hello"
> option, as it still gets a fair number of hits from impatient spammers.
>
> 3. Greylisting. If the sender actually has a valid recipient, but
> they've not sent email to that recipient previously I'll reject for a
> few minutes. Some people reckon that this is not so useful these days,
> but statistically it still appears to knock out a significant number of
> attempts.
>
spam:mail radio reversed overnight and has stayed that way ever since.
> 4. Spamassassin and Antivirus. If you get past everything else, I've got
> some custom rules in SA that mark up particular specific traits. I'm
> also not particularly keen on senders without (decent) rDNS, and tend to
> mark them up significantly.
>
web forum) which gray-listing can't touch and manage to get rid of 90% of
it through SA local rules.
> I've a low volume system for me and my family.
>
> Over the last three
> months I would suggest that greylisting is still the most effective
> deterrent by volume per use of resource.
<
Martin Gregorie
Oct 21, 2011, 7:31:46 PM10/21/11
to
Wayland is not planning to provide remote access any time soon. Speaking
for myself, I rely on using X-term so I can ssh to remote boxes on my
home LAN and, if necessary, run GUI programs there. AFAICT Wayland is not
intending to support this anytime soon, if at all. Read their roadmap: I
did. The relevant buzzword is 'net transparency'.
Nix
Oct 21, 2011, 8:26:41 PM10/21/11
to
On 21 Oct 2011, Nick Leverton told this:
layer in the kernel. No server process. (But, of course, X using a
kernel modesetting X driver *also* talks directly to the DRM layer in
the kernel, as do userspace processes making use of Mesa.)
--
NULL && (void)
> I'd been quite impressed with the idea of Wayland, but I didn't realise
> it runs in kernel space. As a user task it would be quite a good idea.
It doesn't, not really. It's a library that talks directly to the DRM
> it runs in kernel space. As a user task it would be quite a good idea.
layer in the kernel. No server process. (But, of course, X using a
kernel modesetting X driver *also* talks directly to the DRM layer in
the kernel, as do userspace processes making use of Mesa.)
--
NULL && (void)
Aragorn
Oct 22, 2011, 2:29:37 AM10/22/11
to
On Friday 21 October 2011 19:16, Tony Mountifield conveyed the following
to uk.comp.os.linux...
> In article <j7s0mb$j2c$2...@dont-email.me>,
> Aragorn <str...@telenet.be.invalid> wrote:
>
>> On Friday 21 October 2011 16:18, Tony Houghton conveyed the following
>> to uk.comp.os.linux...
>>
>> > Long-term they are planning to replace X with something called
>> > Wayland.
>>
>> ... Which runs in kernelspace, and which is (as yet) not network-
>> transparent, unlike X11.
>
> I remember once upon a time, one of the sticks used for MS-bashing was
> that the Windows graphics subsystem ran in kernel space, which was one
> of the reasons it could be unstable...
My point exactly. ;-)
> In article <j7s0mb$j2c$2...@dont-email.me>,
> Aragorn <str...@telenet.be.invalid> wrote:
>
>> On Friday 21 October 2011 16:18, Tony Houghton conveyed the following
>> to uk.comp.os.linux...
>>
>> > Long-term they are planning to replace X with something called
>> > Wayland.
>>
>> ... Which runs in kernelspace, and which is (as yet) not network-
>> transparent, unlike X11.
>
> I remember once upon a time, one of the sticks used for MS-bashing was
> that the Windows graphics subsystem ran in kernel space, which was one
> of the reasons it could be unstable...
alexd
Oct 22, 2011, 4:07:30 AM10/22/11
to
Martin Gregorie (for it is he) wrote:
> If you're a one-computer sort of guy this may not be a problem, but
> Wayland is not planning to provide remote access any time soon. Speaking
> for myself, I rely on using X-term so I can ssh to remote boxes on my
> home LAN and, if necessary, run GUI programs there.
Are they not planning to support X clients? I can't see how they could
> If you're a one-computer sort of guy this may not be a problem, but
> Wayland is not planning to provide remote access any time soon. Speaking
> for myself, I rely on using X-term so I can ssh to remote boxes on my
> home LAN and, if necessary, run GUI programs there.
manage the migration without support for legacy X clients.
09:04:26 up 35 days, 14:17, 5 users, load average: 0.10, 0.41, 0.33
Message has been deleted
Message has been deleted
Richard Kettlewell
Oct 22, 2011, 4:36:58 AM10/22/11
to
Wayland client.
--
http://www.greenend.org.uk/rjk/
Martin Gregorie
Oct 22, 2011, 8:16:10 AM10/22/11
to
On Sat, 22 Oct 2011 09:07:30 +0100, alexd wrote:
> Martin Gregorie (for it is he) wrote:
>
>> If you're a one-computer sort of guy this may not be a problem, but
>> Wayland is not planning to provide remote access any time soon.
>> Speaking for myself, I rely on using X-term so I can ssh to remote
>> boxes on my home LAN and, if necessary, run GUI programs there.
>
> Are they not planning to support X clients? I can't see how they could
> manage the migration without support for legacy X clients.
>
Provided they expose the same API to the client programs as X11 it should
> Martin Gregorie (for it is he) wrote:
>
>> If you're a one-computer sort of guy this may not be a problem, but
>> Wayland is not planning to provide remote access any time soon.
>> Speaking for myself, I rely on using X-term so I can ssh to remote
>> boxes on my home LAN and, if necessary, run GUI programs there.
>
> Are they not planning to support X clients? I can't see how they could
> manage the migration without support for legacy X clients.
>
work seamlessly. The difference is that, since X11 was designed to
support terminals over a TCP/IP network, there's a network connection
between the API library and the X11 server - on Linux this just happens
to be the localhost simulated/internal network. Wayland would dispense
with this, by gluing the API library directly to the graphics engine.
A useful feature of X11 is that the API library doesn't care where the
server is or who implemented it. Hence the application can talk to the
local Xorg server, an Xorg server on another Linux box across the
network, an X11 server that was implemented on Windows or a Mac or a
physical X11 terminal.
Wayland won't do any of this and, probably won't play nicely with VNC
either, since that also talks to the application via localhost ports.
Martin Gregorie
Oct 22, 2011, 8:21:04 AM10/22/11
to
existing applications, but are they planning to implement OpenGL and/or
DirectX APIs as well?
Tony Houghton
Oct 22, 2011, 9:11:52 AM10/22/11
to
In <j7ucfg$nsu$1...@localhost.localdomain>,
Martin Gregorie <mar...@address-in-sig.invalid> wrote:
[Wayland]
> Presumably they'll have to provide the X11 APIs to avoid breaking
> existing applications,
I think they're planning new versions of GTK+ and Qt which should take
care of most applications (but some will have to disable X hacks used to
work around issues like
<https://bugzilla.gnome.org/show_bug.cgi?id=586664>). And there'll be
that X compatibility layer mentioned elsewhere in the thread.
> but are they planning to implement OpenGL and/or
> DirectX APIs as well?
I can't imagine there won't be OpenGL support. I didn't think there
would be DirectX until I found
<http://www.phoronix.com/scan.php?page=article&item=mesa_gallium3d_d3d11&num=1>,
but there's a lot more to full DirectX than Direct3D.
Martin Gregorie <mar...@address-in-sig.invalid> wrote:
[Wayland]
> Presumably they'll have to provide the X11 APIs to avoid breaking
> existing applications,
care of most applications (but some will have to disable X hacks used to
work around issues like
<https://bugzilla.gnome.org/show_bug.cgi?id=586664>). And there'll be
that X compatibility layer mentioned elsewhere in the thread.
> but are they planning to implement OpenGL and/or
> DirectX APIs as well?
would be DirectX until I found
<http://www.phoronix.com/scan.php?page=article&item=mesa_gallium3d_d3d11&num=1>,
but there's a lot more to full DirectX than Direct3D.
Nix
Oct 22, 2011, 9:23:09 AM10/22/11
to
On 22 Oct 2011, Martin Gregorie uttered the following:
networky stuff if necesasry. (In theory, less efficiently than X11,
since it'd have to throw bitmaps about... but modern apps are just
throwing bitmaps about *anyway*: almost nobody uses the core X drawing
API anymore. So the actual efficiency loss seems likely to be minimal,
and possibly an improvement, given that the protocol can be written from
scratch to minimize roundtrips.)
--
NULL && (void)
> Wayland won't do any of this and, probably won't play nicely with VNC
> either, since that also talks to the application via localhost ports.
However, it does require a compositor, so the compositor can do lots of
> either, since that also talks to the application via localhost ports.
networky stuff if necesasry. (In theory, less efficiently than X11,
since it'd have to throw bitmaps about... but modern apps are just
throwing bitmaps about *anyway*: almost nobody uses the core X drawing
API anymore. So the actual efficiency loss seems likely to be minimal,
and possibly an improvement, given that the protocol can be written from
scratch to minimize roundtrips.)
--
NULL && (void)
Nix
Oct 22, 2011, 9:24:54 AM10/22/11
to
On 22 Oct 2011, Martin Gregorie spake thusly:
> On Sat, 22 Oct 2011 09:36:58 +0100, Richard Kettlewell wrote:
>> Their FAQ addresses this, the answer being an X server running as a
>> Wayland client.
>
> Presumably they'll have to provide the X11 APIs to avoid breaking
> existing applications,
There is no plan for this, just an embedded X server running atop
Wayland (hopefully rootlessly, so X windows can interoperate with
Wayland ones). Toolkits will have to implement a Wayland backend,
but there aren't many active toolkits now and most of them already have
one under development.
> but are they planning to implement OpenGL and/or
> DirectX APIs as well?
Mesa is still part of the stack, so OpenGL is already provided, and
Gallium (now folded into Mesa) has a DirectX API layer under
(intermittent) development too.
--
NULL && (void)
> On Sat, 22 Oct 2011 09:36:58 +0100, Richard Kettlewell wrote:
>> Their FAQ addresses this, the answer being an X server running as a
>> Wayland client.
>
> Presumably they'll have to provide the X11 APIs to avoid breaking
> existing applications,
Wayland (hopefully rootlessly, so X windows can interoperate with
Wayland ones). Toolkits will have to implement a Wayland backend,
but there aren't many active toolkits now and most of them already have
one under development.
> but are they planning to implement OpenGL and/or
> DirectX APIs as well?
Gallium (now folded into Mesa) has a DirectX API layer under
(intermittent) development too.
--
NULL && (void)
Martin Gregorie
Oct 22, 2011, 11:44:56 AM10/22/11
to
On Sat, 22 Oct 2011 13:11:52 +0000, Tony Houghton wrote:
> I can't imagine there won't be OpenGL support. I didn't think there
> would be DirectX until I found
> <http://www.phoronix.com/scan.php?
page=article&item=mesa_gallium3d_d3d11&num=1>,
> but there's a lot more to full DirectX than Direct3D.
Very interesting indeed. I wonder if the Wine people are aware of
> I can't imagine there won't be OpenGL support. I didn't think there
> would be DirectX until I found
> <http://www.phoronix.com/scan.php?
page=article&item=mesa_gallium3d_d3d11&num=1>,
> but there's a lot more to full DirectX than Direct3D.
Mesa (or Wayland for that matter).
alexd
Oct 22, 2011, 1:51:21 PM10/22/11
to
Martin Gregorie (for it is he) wrote:
> Provided they expose the same API to the client programs as X11 it should
> work seamlessly. The difference is that, since X11 was designed to
> support terminals over a TCP/IP network, there's a network connection
> between the API library and the X11 server - on Linux this just happens
> to be the localhost simulated/internal network. Wayland would dispense
> with this, by gluing the API library directly to the graphics engine.
Either they provide an X implementation or compatible server, or they don't.
> Provided they expose the same API to the client programs as X11 it should
> work seamlessly. The difference is that, since X11 was designed to
> support terminals over a TCP/IP network, there's a network connection
> between the API library and the X11 server - on Linux this just happens
> to be the localhost simulated/internal network. Wayland would dispense
> with this, by gluing the API library directly to the graphics engine.
They can't expose the same API, but not support networking.
18:49:39 up 36 days, 3 min, 5 users, load average: 0.40, 1.27, 0.89
Martin Gregorie
Oct 22, 2011, 6:08:45 PM10/22/11
to
On Sat, 22 Oct 2011 18:51:21 +0100, alexd wrote:
> Martin Gregorie (for it is he) wrote:
>
>> Provided they expose the same API to the client programs as X11 it
>> should work seamlessly. The difference is that, since X11 was designed
>> to support terminals over a TCP/IP network, there's a network
>> connection between the API library and the X11 server - on Linux this
>> just happens to be the localhost simulated/internal network. Wayland
>> would dispense with this, by gluing the API library directly to the
>> graphics engine.
>
> Either they provide an X implementation or compatible server, or they
> don't. They can't expose the same API, but not support networking.
>
How do you work that out?
> Martin Gregorie (for it is he) wrote:
>
>> Provided they expose the same API to the client programs as X11 it
>> should work seamlessly. The difference is that, since X11 was designed
>> to support terminals over a TCP/IP network, there's a network
>> connection between the API library and the X11 server - on Linux this
>> just happens to be the localhost simulated/internal network. Wayland
>> would dispense with this, by gluing the API library directly to the
>> graphics engine.
>
> Either they provide an X implementation or compatible server, or they
> don't. They can't expose the same API, but not support networking.
>
There are a whole stack of data structures and function prototypes
defined in X11/Xlib.h and a collection of matching functions implemented
in Xlib. From the programmer's POV these are just another set of library
procedures that you call to drive a graphical display and read keyboard
and mouse inputs. It doesn't and must not matter to a programmer whether
they use system calls to drive Wayland or a network connection to talk to
an X.11 server.
In fact even the programming API could be agnostic about the server
interface. If the Xlib library equivalent is a two layer implementation
the program would always bind to the same API layer and that in turn
would determine what it type of graphics engine it is going to talk to
and where it is before loading the appropriate second layer module, which
would be graphics engine and request protocol specific.
If it isn't like that, Wayland would be DOA because the last thing
anybody wants is to be unable to point a program at X.11 one day and at
Wayland the next without recompiling it.
Nix
Oct 22, 2011, 7:07:09 PM10/22/11
to
On 22 Oct 2011, Martin Gregorie spake thusly:
> Very interesting indeed. I wonder if the Wine people are aware of
> Mesa
That's contentious. Luca thinks so:
> Mesa
<http://www.winehq.org/pipermail/wine-devel/2010-September/086908.html>.
The Wine devs disagree:
<http://www.winehq.org/pipermail/wine-devel/2010-September/086901.html>.
(This doesn't make those drivers useless, as Direct3D is a much nicer
API in some ways than OpenGL, and it would be nice to have a choice.)
--
NULL && (void)
Richard Kettlewell
Oct 23, 2011, 6:07:20 AM10/23/11
to
Martin Gregorie <mar...@address-in-sig.invalid> writes:
client library.
> In fact even the programming API could be agnostic about the server
> interface. If the Xlib library equivalent is a two layer
> implementation the program would always bind to the same API layer and
> that in turn would determine what it type of graphics engine it is
> going to talk to and where it is before loading the appropriate second
> layer module, which would be graphics engine and request protocol
> specific.
I suggest reading the Wayland FAQ. They talk about X11 servers as
clients of Wayland. They don't talk about reimplementing Xlib. That
would be stupid in any case.
Reimplementing GDK on top of Wayland, now that would actually make some
sense. And oh look, someone's done it.
> If it isn't like that, Wayland would be DOA because the last thing
> anybody wants is to be unable to point a program at X.11 one day and
> at Wayland the next without recompiling it.
You point your X11-only clients at an X11 server that uses Wayland to
draw that in case. I'm currently using a system that works exactly this
way (but with a different underlying display API to Wayland).
--
http://www.greenend.org.uk/rjk/
> alexd wrote:
>> Either they provide an X implementation or compatible server, or they
>> don't. They can't expose the same API, but not support networking.
>
> How do you work that out?
>
> There are a whole stack of data structures and function prototypes
> defined in X11/Xlib.h and a collection of matching functions
> implemented in Xlib. From the programmer's POV these are just another
> set of library procedures that you call to drive a graphical display
> and read keyboard and mouse inputs. It doesn't and must not matter to
> a programmer whether they use system calls to drive Wayland or a
> network connection to talk to an X.11 server.
Xlib isn't the only X11 C API. X11 is the protocol not a particular
>> Either they provide an X implementation or compatible server, or they
>> don't. They can't expose the same API, but not support networking.
>
> How do you work that out?
>
> There are a whole stack of data structures and function prototypes
> defined in X11/Xlib.h and a collection of matching functions
> implemented in Xlib. From the programmer's POV these are just another
> set of library procedures that you call to drive a graphical display
> and read keyboard and mouse inputs. It doesn't and must not matter to
> a programmer whether they use system calls to drive Wayland or a
> network connection to talk to an X.11 server.
client library.
> In fact even the programming API could be agnostic about the server
> interface. If the Xlib library equivalent is a two layer
> implementation the program would always bind to the same API layer and
> that in turn would determine what it type of graphics engine it is
> going to talk to and where it is before loading the appropriate second
> layer module, which would be graphics engine and request protocol
> specific.
clients of Wayland. They don't talk about reimplementing Xlib. That
would be stupid in any case.
Reimplementing GDK on top of Wayland, now that would actually make some
sense. And oh look, someone's done it.
> If it isn't like that, Wayland would be DOA because the last thing
> anybody wants is to be unable to point a program at X.11 one day and
> at Wayland the next without recompiling it.
draw that in case. I'm currently using a system that works exactly this
way (but with a different underlying display API to Wayland).
--
http://www.greenend.org.uk/rjk/
Nigel Wade
Oct 24, 2011, 5:12:35 AM10/24/11
to
On 22/10/11 00:08, Martin Gregorie wrote:
> On Thu, 20 Oct 2011 16:52:58 +0200, A. Filip wrote:
>> I use periodic fetchmail polling but * you may consider using IMAP IDLE
>> for getting messages with negligible
>>
> I used to use it, then got pissed off by the way it left read-but-not-
> deleted mail in my ISP mailbox forever or I manually deleted it. Since
> then I switched to using getmail, which 'just works' and even uses the
> same mda chain. Recommended. my distro, Fedora, includes it as a standard
> package. Yours may too.
>
>
But that's exactly the point of IMAP. That your mail remains on the
> On Thu, 20 Oct 2011 16:52:58 +0200, A. Filip wrote:
>> I use periodic fetchmail polling but * you may consider using IMAP IDLE
>> for getting messages with negligible
>>
> I used to use it, then got pissed off by the way it left read-but-not-
> deleted mail in my ISP mailbox forever or I manually deleted it. Since
> then I switched to using getmail, which 'just works' and even uses the
> same mda chain. Recommended. my distro, Fedora, includes it as a standard
> package. Yours may too.
>
>
server and is accessible from any client, not just the one you happened
to connect from, and download it to.
--
Nigel Wade
Justin C
Oct 24, 2011, 9:18:32 AM10/24/11
to
On 2011-10-21, A. Filip <an...@onet.eu> wrote:
> Justin C <justi...@purestblue.com> wrote:
>> On 2011-10-21, Mike Scott <usen...@scottsonline.org.uk.invalid> wrote:
>>
>> [snip]
>>>
>>> Or use your own domain name. Use dyndns so you have a handle to your
>>> mail server, and set up the MX records for your domain accordingly.
>>> Works quite well enough for a domestic system on a dynamic ip.
>>>
>>> Side effect - you can lock out an awful lot of spammers with your own
>>> firewall.
>>
>> I'd be interested to know how you do this. I host mail for my own
>> domain and Exim handles most of the spam rejection, if I can reject it
>> at the firewall then I'd like to do that.
>
> As I can see it (simple/static) Linux firewall rules can:
> a) block connections from nets with highest zombies infection rate
> b) reduce system load by implementing "cheap" incoming SMTP connections
> rate limits (may work well with grey-listing)
>
> Spamhaus provides lists of nets "recommended for firewalling out"
> http://www.spamhaus.org/drop/
Thanks, I will investigate.
> Justin C <justi...@purestblue.com> wrote:
>> On 2011-10-21, Mike Scott <usen...@scottsonline.org.uk.invalid> wrote:
>>
>> [snip]
>>>
>>> Or use your own domain name. Use dyndns so you have a handle to your
>>> mail server, and set up the MX records for your domain accordingly.
>>> Works quite well enough for a domestic system on a dynamic ip.
>>>
>>> Side effect - you can lock out an awful lot of spammers with your own
>>> firewall.
>>
>> I'd be interested to know how you do this. I host mail for my own
>> domain and Exim handles most of the spam rejection, if I can reject it
>> at the firewall then I'd like to do that.
>
> As I can see it (simple/static) Linux firewall rules can:
> a) block connections from nets with highest zombies infection rate
> b) reduce system load by implementing "cheap" incoming SMTP connections
> rate limits (may work well with grey-listing)
>
> Spamhaus provides lists of nets "recommended for firewalling out"
> http://www.spamhaus.org/drop/
Justin.
--
Justin C, by the sea.
Martin Gregorie
Oct 24, 2011, 6:49:36 PM10/24/11
to
On Mon, 24 Oct 2011 10:12:35 +0100, Nigel Wade wrote:
> But that's exactly the point of IMAP. That your mail remains on the
> server and is accessible from any client, not just the one you happened
> to connect from, and download it to.
>
You misunderstand. My problem was to reliably vacuum all the mail in my
> But that's exactly the point of IMAP. That your mail remains on the
> server and is accessible from any client, not just the one you happened
> to connect from, and download it to.
>
ISP's delivery box onto my system and not leave anything behind,
especially not copies of mail that I'd already sucked down. fetchmail
proved to have bugs that stop it doing this reliably. getmail doesn't.
In this case IMAP is irrelevant. I have no need to leave any mail on my
ISP's server. Instead I keep copies of everything on my own mailserver
where its behind my firewall and under my control. Old mail is in my
PostgreSQL- based archival system which I can search faster than any
mailfolder based system I've every seen.
IOW, If I saw any advantage in using IMAP, the mailfolders would be on my
own server and accessible via Dovecot. They would definitely NOT be on my
ISP's kit.
A. Filip
Oct 25, 2011, 2:04:13 AM10/25/11
to
Martin Gregorie <mar...@address-in-sig.invalid> wrote:
> On Thu, 20 Oct 2011 16:52:58 +0200, A. Filip wrote:
>> I use periodic fetchmail polling but * you may consider using IMAP IDLE
>> for getting messages with negligible
>>
> I used to use it, then got pissed off by the way it left read-but-not-
> deleted mail in my ISP mailbox forever or I manually deleted it. Since
> then I switched to using getmail, which 'just works' and even uses the
> same mda chain. Recommended. my distro, Fedora, includes it as a standard
> package. Yours may too.
I think I can remember similar problems (very) long time ago.
> On Thu, 20 Oct 2011 16:52:58 +0200, A. Filip wrote:
>> I use periodic fetchmail polling but * you may consider using IMAP IDLE
>> for getting messages with negligible
>>
> I used to use it, then got pissed off by the way it left read-but-not-
> deleted mail in my ISP mailbox forever or I manually deleted it. Since
> then I switched to using getmail, which 'just works' and even uses the
> same mda chain. Recommended. my distro, Fedora, includes it as a standard
> package. Yours may too.
I think I can not remember anything that could not be solved/fixed by
using --all command line option during a few fetches.
--
[pl>en Andrew] Andrzej A. Filip : an...@onet.eu : Andrze...@gmail.com
I found Rome a city of bricks and left it a city of marble.
-- Augustus Caesar
Nigel Wade
Oct 25, 2011, 4:46:28 AM10/25/11
to
On 24/10/11 23:49, Martin Gregorie wrote:
> On Mon, 24 Oct 2011 10:12:35 +0100, Nigel Wade wrote:
>
>> But that's exactly the point of IMAP. That your mail remains on the
>> server and is accessible from any client, not just the one you happened
>> to connect from, and download it to.
>>
> You misunderstand. My problem was to reliably vacuum all the mail in my
> ISP's delivery box onto my system and not leave anything behind,
> especially not copies of mail that I'd already sucked down. fetchmail
> proved to have bugs that stop it doing this reliably. getmail doesn't.
>
Yes, I misunderstood. I thought you were complaining about IMAP, not
> On Mon, 24 Oct 2011 10:12:35 +0100, Nigel Wade wrote:
>
>> But that's exactly the point of IMAP. That your mail remains on the
>> server and is accessible from any client, not just the one you happened
>> to connect from, and download it to.
>>
> You misunderstand. My problem was to reliably vacuum all the mail in my
> ISP's delivery box onto my system and not leave anything behind,
> especially not copies of mail that I'd already sucked down. fetchmail
> proved to have bugs that stop it doing this reliably. getmail doesn't.
>
fetchmail.
--
Nigel Wade
Johny B Good
Oct 28, 2011, 8:23:01 AM10/28/11
to
On Fri, 21 Oct 2011 15:33:15 +0100, Nix <nix-ra...@esperi.org.uk>
wrote:
> On 21 Oct 2011, Johny B. Good outgrape:
>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>> takes between 46 and 52 watts when idle (no spin down power saving
>> involved) depending on what 'mood' the two 'so called eco green'
>> Western Digital WD20EARS drives[2] happen to be in.
First off, sorry for the late response. Secondly, oops! When I used the
phrase "jbod, not RAID", I was mis-using it as shorthand for separate disk
volumes on the server box rather than a single gigantic 8(short)TB disk
volume.
>
> These drives do spin down to save power (I know, I have four WD10EACS
> drives in my RAID array, half the size of yours and still nowhere near
> full). They don't spin all the way down, though, just to about 2000rpm,
> then back up if I/O becomes frequent enough. They keep servicing
> requests the whole time. Sodding awesome transfer rate with four bound
> together over PCIe too: between 190MiB/s at high sectors up to 250MiB/s
> at low. My previous aged PCI software RAID array could manage 30Mb/s if
> you were very lucky...
Oh, how things have changed!
>
> (The machine itself is not ancient but a two-year-old headless Nehalem
> server box, 24Gb RAM and all: I do anything that requires serious
> gruntwork on it.)
Mine is about that age too. It's only the case and its PSU that's
"ancient".
>
>> [1] You might think that an old small form factor 145W rated ATX PSU
>> might succumb to overload on its 4.2A rated 12v rail with all those
>> hard disk drives but it takes the spin up loading in its stride (even
>
> Yeah, four disks are unlikely to cause problems. Eight might though. (My
> RAID array is a hardware one and has an option to stagger spinup
> specifically to avoid overloading the PSU.)
That's a given, especially true of SCSI arrays.
>
>> I stuck with the original PSU on account of its better efficiency and
>> thermally regulated fan which provided all of the (very quiet) cooling
>> required of the whole box.
>
> One would hope newer PSUs also have fans that don't spin at top speed
> the whole time, given that they sound like jet engines when they do.
Well, if the thermal speed control built into a cheap Octigen ATX PSU I
cannibalised recently to repair the PSU in my desktop machine is anything
to go by, then I'm not surprised they suffer short lives from overheating.
In this case, the slow speed setting was far too slow and the thermal cut
in point was set so high the top of the tower became very warm to the
touch (causing the CPU fan to resort to full speed to compensate for the
very high temps never ever seen before with my previous PSU setup).
The PSU itself was almost too hot to touch and I landed up disabling the
built in thermal control and transplanting my homebrewed thermal module
I'd made for the previous, non regulated PSU PCB I'd used in order to get
back to a decent base level (very quiet) airflow to drop the temps back
down to sane levels.
Regarding the 145W unit in the Gateway 2000 case, I've never heard it
speed up (except possibly during heatwave conditions a few years back when
the room temperature got above the 27 dec C mark) and this very nicely
keeps all four drives below the 13 deg above ambient mark.
>
>> [2] I've used the WDIDLE3 utility to increase the head unloading
>> timeout from the egregiously short 8.3 seconds to a more reasonable 5
>> minutes
>
> Interesting. The heads don't unload on mine at all: this is even touted
> as a feature ('cos it means you can always get at your data with zero
> delay). Maybe yours are laptop drives or something?
>
>> (you can't actually totally disable this feature on these
>> drives). I prefer not to use spindown timeout power saving on hard
>> disk drives,
>
> I do it simply because it saves me power and thus money. It is also
> reasonable to assume that the drives are expected to operate mostly at
> the lower speed, so continuous high-speed spinning might shorten their
> life.
>
>> The Western Digitals were my second choice in the 2TB eco green drive
>> short list and were acquired to satisfy a critical capacity upgrade
>> otherwise I'd have only used Samsungs. Now, of course, Samsung is no
>> longer an option in the 3TB and larger eco green drive capacities that
>> I'm now having to consider for the next capacity upgrade. I'm hoping
>> against hope that the next Western Digital eco green drives will be
>> much better behaved than those WD20EARS units.
>
> I see no behaviour problems with my WD10EACS. What misbehaviour do you
> see? Does it actually stop sending you data and spin all the way down
> after only 8.3s? That seems very unlikely: are you sure you didn't
> misinterpret the spindown-to-2k-RPM setting as a complete spindown?
No, not spindown. The issue isn't the very slight response delay so much
as the 300,000 head unload life limit rating and the fact that, with the
unconscionable default 8 second time-out, this can be reached in as little
as 6 months or less of use.
As a matter of fact (and I can't work out why, considering the load cycle
count of 9572 on the second unit, commissioned only just 85 hours later)
the first of the WD20EARS drives now shows a load cycle count figure of
366542 (up from the 366539 figure noted just 21 hours ago). In view of the
low accumulation rate that now seems to apply, I can only assume I didn't
manage to apply the WDIDLE3 fix until a few months after I'd managed to
apply it to the second drive.
If the head unload rating life rating was 3,000,000, it wouldn't be an
issue but, even taking WD's revised estimate of 600,000 into account, this
still represents a possible failure within a mere 12 months of use so it's
vital that the time-out be raised to something more reasonable (like 300
seconds!) if you want to give the drive a fighting chance to survive the 3
(5?) year warranty period.
However, head unloading time out issues aside, it's the strange non
eco-green power consumption curve that irks me the most. When I installed
those drives, I used them to replace a couple of 7200 rpm 1TB Samsung
spinpoints that lived alongside a couple of 1TB Samsung 5400 rpm eco green
drives, expecting the 53 watt consumption to fall below the 49 watt mark
but all that happened was that the power varied between 48 and 53 watts
(infrequently as low as 48, most often 53 watts regardless of how long the
box was effectively 'idle').
The drive temperature observations correlated with these variations,
proving the WDs to be the culprits in this random see-sawing of power
consumption. Previous "mystery power variations" with the earlier set of
72000 rpm 1TB Spinpoints had eventually proved to be a (very consistent
and predictable) drive (fluid dynamic) bearing lubricant viscosity
temperature dependent effect. A temperature drop from 20 to 16 deg C
causing the wattage to increase by 2 watts due to the increased spindle
drag on all four SpinPoint drives fitted in the box. In short, the
eco-green credentials of those WD20EARS drives are very questionable IME.
As it happens, I've just replaced one of the two Samsung eco-green drives
with an Hitachi 3TB 5k3000 Coolspin drive (supposedly an eco-green drive
with impressively low power consumption (spinning) figures) and this has
added yet a couple more watts consumption.
I would have replaced one of those pesky WDs but I needed the 512 byte
sectored[1] 2TB Samsung to replace a 1TB 7200 rpm Samsung in my win2k box
so that I could borrow its controller board to retrieve customer data off
of another Samsung 7200 rpm drive that had died just after wiping the
original 250GB Seagate that had been successfully cloned onto it so it
could be recycled for relocating the pagefile and users' data off of the
new boot drive.
As it happened, the controller PCB transplant trick worked just fine and
I successfully retrieved all the data. The problem was that I
under-estimated the PCB's susceptibility to ESD[2] when it came to
refitting it back to the original drive it had been borrowed from and I
now have two dead Samsung drives with perfect mechanicals for my trouble.
Having one drive, randomly decide to 'let out the magic smoke' was bad
enough, having another one do likewise due to carelessness on my part is a
rather bitter pill to swallow. I'm hanging onto those units in the hope
that I'll see a customer (or two) bring in a 1TB 7200 rpm Spinpoint with
bad sectors (rather than a dead controller). Like the lottery, "You've got
to be in it to win it!", I need to keep hold of them to stand any chance
of being so lucky. Much better odds than the lottery but still almost non
existent. You just never know and, besides, I don't want to give Murphy
any chance to rub my nose in it by immediately chucking them away.
[1] What I'd initially assumed was a 4k sector issue with win2k (even
after correctly aligning the partition boundary) looks like it may have
had more to do with an e-SATA issue within the win2k box since I saw
exactly the same slow and stuttered write performance along with mouse
movement hangs when I tested the newly released and re-partitioned 512
byte sectored 2TB eco-green Samsung drive in an e-SATA enclosure.
The MoBo I'm using for the win2k box is blessed with a couple of e-SATA
ports in the IO shield area but these rely on the use of sata data cables
to link to the two designated port headers (strangely ports 1 & 2, rather
than 3 & 4) of the 4 internal SATA ports and it seems (amply demonstrating
the flakiness of SATA data connectors) that a poor link connection was the
true cause of the problem. I removed the case cover and directly connected
the drive, bypassing the e-SATA data link connections and was able to get
full speed data transfers. I've since replaced the link cable with a
better(???) sata data cable and the external e-SATA drives are back to
normal again.
[2] In this case, re-uniting two items (never ever expected to be
separated outside of the manufacturers' facilities) that lacked any ESD
"hardening" on their 'permanent' interface connections (specifically the
read/write head connections) where the every day ESD handling precautions
become insufficient.
Please regard this as a cautionary tale for those wishing to use the
"swap out controller PCBs" method of data recovery. The issue isn't that
this won't work but that of the need to be ultra protective against the
risk of ESD induced damage to the controller board. In this case, I was
lucky it didn't happen until _after_ I'd recovered the data - it could
have happened when joining the board to the 'dead' drive and the exercise
would not have had any up side to compensate the unintended sacrifice of
an otherwise perfectly good disk drive.
--
Regards JB Good
wrote:
> On 21 Oct 2011, Johny B. Good outgrape:
>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>> takes between 46 and 52 watts when idle (no spin down power saving
>> involved) depending on what 'mood' the two 'so called eco green'
>> Western Digital WD20EARS drives[2] happen to be in.
First off, sorry for the late response. Secondly, oops! When I used the
phrase "jbod, not RAID", I was mis-using it as shorthand for separate disk
volumes on the server box rather than a single gigantic 8(short)TB disk
volume.
>
> These drives do spin down to save power (I know, I have four WD10EACS
> drives in my RAID array, half the size of yours and still nowhere near
> full). They don't spin all the way down, though, just to about 2000rpm,
> then back up if I/O becomes frequent enough. They keep servicing
> requests the whole time. Sodding awesome transfer rate with four bound
> together over PCIe too: between 190MiB/s at high sectors up to 250MiB/s
> at low. My previous aged PCI software RAID array could manage 30Mb/s if
> you were very lucky...
Oh, how things have changed!
>
> (The machine itself is not ancient but a two-year-old headless Nehalem
> server box, 24Gb RAM and all: I do anything that requires serious
> gruntwork on it.)
Mine is about that age too. It's only the case and its PSU that's
"ancient".
>
>> [1] You might think that an old small form factor 145W rated ATX PSU
>> might succumb to overload on its 4.2A rated 12v rail with all those
>> hard disk drives but it takes the spin up loading in its stride (even
>
> Yeah, four disks are unlikely to cause problems. Eight might though. (My
> RAID array is a hardware one and has an option to stagger spinup
> specifically to avoid overloading the PSU.)
That's a given, especially true of SCSI arrays.
>
>> I stuck with the original PSU on account of its better efficiency and
>> thermally regulated fan which provided all of the (very quiet) cooling
>> required of the whole box.
>
> One would hope newer PSUs also have fans that don't spin at top speed
> the whole time, given that they sound like jet engines when they do.
Well, if the thermal speed control built into a cheap Octigen ATX PSU I
cannibalised recently to repair the PSU in my desktop machine is anything
to go by, then I'm not surprised they suffer short lives from overheating.
In this case, the slow speed setting was far too slow and the thermal cut
in point was set so high the top of the tower became very warm to the
touch (causing the CPU fan to resort to full speed to compensate for the
very high temps never ever seen before with my previous PSU setup).
The PSU itself was almost too hot to touch and I landed up disabling the
built in thermal control and transplanting my homebrewed thermal module
I'd made for the previous, non regulated PSU PCB I'd used in order to get
back to a decent base level (very quiet) airflow to drop the temps back
down to sane levels.
Regarding the 145W unit in the Gateway 2000 case, I've never heard it
speed up (except possibly during heatwave conditions a few years back when
the room temperature got above the 27 dec C mark) and this very nicely
keeps all four drives below the 13 deg above ambient mark.
>
>> [2] I've used the WDIDLE3 utility to increase the head unloading
>> timeout from the egregiously short 8.3 seconds to a more reasonable 5
>> minutes
>
> Interesting. The heads don't unload on mine at all: this is even touted
> as a feature ('cos it means you can always get at your data with zero
> delay). Maybe yours are laptop drives or something?
>
>> (you can't actually totally disable this feature on these
>> drives). I prefer not to use spindown timeout power saving on hard
>> disk drives,
>
> I do it simply because it saves me power and thus money. It is also
> reasonable to assume that the drives are expected to operate mostly at
> the lower speed, so continuous high-speed spinning might shorten their
> life.
>
>> The Western Digitals were my second choice in the 2TB eco green drive
>> short list and were acquired to satisfy a critical capacity upgrade
>> otherwise I'd have only used Samsungs. Now, of course, Samsung is no
>> longer an option in the 3TB and larger eco green drive capacities that
>> I'm now having to consider for the next capacity upgrade. I'm hoping
>> against hope that the next Western Digital eco green drives will be
>> much better behaved than those WD20EARS units.
>
> I see no behaviour problems with my WD10EACS. What misbehaviour do you
> see? Does it actually stop sending you data and spin all the way down
> after only 8.3s? That seems very unlikely: are you sure you didn't
> misinterpret the spindown-to-2k-RPM setting as a complete spindown?
No, not spindown. The issue isn't the very slight response delay so much
as the 300,000 head unload life limit rating and the fact that, with the
unconscionable default 8 second time-out, this can be reached in as little
as 6 months or less of use.
As a matter of fact (and I can't work out why, considering the load cycle
count of 9572 on the second unit, commissioned only just 85 hours later)
the first of the WD20EARS drives now shows a load cycle count figure of
366542 (up from the 366539 figure noted just 21 hours ago). In view of the
low accumulation rate that now seems to apply, I can only assume I didn't
manage to apply the WDIDLE3 fix until a few months after I'd managed to
apply it to the second drive.
If the head unload rating life rating was 3,000,000, it wouldn't be an
issue but, even taking WD's revised estimate of 600,000 into account, this
still represents a possible failure within a mere 12 months of use so it's
vital that the time-out be raised to something more reasonable (like 300
seconds!) if you want to give the drive a fighting chance to survive the 3
(5?) year warranty period.
However, head unloading time out issues aside, it's the strange non
eco-green power consumption curve that irks me the most. When I installed
those drives, I used them to replace a couple of 7200 rpm 1TB Samsung
spinpoints that lived alongside a couple of 1TB Samsung 5400 rpm eco green
drives, expecting the 53 watt consumption to fall below the 49 watt mark
but all that happened was that the power varied between 48 and 53 watts
(infrequently as low as 48, most often 53 watts regardless of how long the
box was effectively 'idle').
The drive temperature observations correlated with these variations,
proving the WDs to be the culprits in this random see-sawing of power
consumption. Previous "mystery power variations" with the earlier set of
72000 rpm 1TB Spinpoints had eventually proved to be a (very consistent
and predictable) drive (fluid dynamic) bearing lubricant viscosity
temperature dependent effect. A temperature drop from 20 to 16 deg C
causing the wattage to increase by 2 watts due to the increased spindle
drag on all four SpinPoint drives fitted in the box. In short, the
eco-green credentials of those WD20EARS drives are very questionable IME.
As it happens, I've just replaced one of the two Samsung eco-green drives
with an Hitachi 3TB 5k3000 Coolspin drive (supposedly an eco-green drive
with impressively low power consumption (spinning) figures) and this has
added yet a couple more watts consumption.
I would have replaced one of those pesky WDs but I needed the 512 byte
sectored[1] 2TB Samsung to replace a 1TB 7200 rpm Samsung in my win2k box
so that I could borrow its controller board to retrieve customer data off
of another Samsung 7200 rpm drive that had died just after wiping the
original 250GB Seagate that had been successfully cloned onto it so it
could be recycled for relocating the pagefile and users' data off of the
new boot drive.
As it happened, the controller PCB transplant trick worked just fine and
I successfully retrieved all the data. The problem was that I
under-estimated the PCB's susceptibility to ESD[2] when it came to
refitting it back to the original drive it had been borrowed from and I
now have two dead Samsung drives with perfect mechanicals for my trouble.
Having one drive, randomly decide to 'let out the magic smoke' was bad
enough, having another one do likewise due to carelessness on my part is a
rather bitter pill to swallow. I'm hanging onto those units in the hope
that I'll see a customer (or two) bring in a 1TB 7200 rpm Spinpoint with
bad sectors (rather than a dead controller). Like the lottery, "You've got
to be in it to win it!", I need to keep hold of them to stand any chance
of being so lucky. Much better odds than the lottery but still almost non
existent. You just never know and, besides, I don't want to give Murphy
any chance to rub my nose in it by immediately chucking them away.
[1] What I'd initially assumed was a 4k sector issue with win2k (even
after correctly aligning the partition boundary) looks like it may have
had more to do with an e-SATA issue within the win2k box since I saw
exactly the same slow and stuttered write performance along with mouse
movement hangs when I tested the newly released and re-partitioned 512
byte sectored 2TB eco-green Samsung drive in an e-SATA enclosure.
The MoBo I'm using for the win2k box is blessed with a couple of e-SATA
ports in the IO shield area but these rely on the use of sata data cables
to link to the two designated port headers (strangely ports 1 & 2, rather
than 3 & 4) of the 4 internal SATA ports and it seems (amply demonstrating
the flakiness of SATA data connectors) that a poor link connection was the
true cause of the problem. I removed the case cover and directly connected
the drive, bypassing the e-SATA data link connections and was able to get
full speed data transfers. I've since replaced the link cable with a
better(???) sata data cable and the external e-SATA drives are back to
normal again.
[2] In this case, re-uniting two items (never ever expected to be
separated outside of the manufacturers' facilities) that lacked any ESD
"hardening" on their 'permanent' interface connections (specifically the
read/write head connections) where the every day ESD handling precautions
become insufficient.
Please regard this as a cautionary tale for those wishing to use the
"swap out controller PCBs" method of data recovery. The issue isn't that
this won't work but that of the need to be ultra protective against the
risk of ESD induced damage to the controller board. In this case, I was
lucky it didn't happen until _after_ I'd recovered the data - it could
have happened when joining the board to the 'dead' drive and the exercise
would not have had any up side to compensate the unintended sacrifice of
an otherwise perfectly good disk drive.
--
Regards JB Good
Nix
Oct 28, 2011, 10:31:03 AM10/28/11
to
On 28 Oct 2011, Johny B. Good stated:
> On Fri, 21 Oct 2011 15:33:15 +0100, Nix <nix-ra...@esperi.org.uk> wrote:
>
>> On 21 Oct 2011, Johny B. Good outgrape:
>>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>>> takes between 46 and 52 watts when idle (no spin down power saving
>>> involved) depending on what 'mood' the two 'so called eco green'
>>> Western Digital WD20EARS drives[2] happen to be in.
>
> First off, sorry for the late response. Secondly, oops! When I used
> the phrase "jbod, not RAID", I was mis-using it as shorthand for
> separate disk volumes on the server box rather than a single gigantic
> 8(short)TB disk volume.
That *is* JBOD, 'just a bunch of disks', not even RAID-0.
>> at low. My previous aged PCI software RAID array could manage 30Mb/s if
>> you were very lucky...
>
> Oh, how things have changed!
I think we can probably blame the sym53c875 component of the array for
that, and the multiple layers of converters hanging off it to allow it
to connect to anything (welcome to SCSI hell).
>>> [1] You might think that an old small form factor 145W rated ATX PSU
>>> might succumb to overload on its 4.2A rated 12v rail with all those
>>> hard disk drives but it takes the spin up loading in its stride (even
>>
>> Yeah, four disks are unlikely to cause problems. Eight might though. (My
>> RAID array is a hardware one and has an option to stagger spinup
>> specifically to avoid overloading the PSU.)
>
> That's a given, especially true of SCSI arrays.
Mine's not SCSI, just normal AHCI (well, it *appears* as normal AHCI to
the drives and to the system as a whole, though obviously normal AHCI
controllers don't have a gig of cache and their own fan, web server, and
SNMP support).
>> One would hope newer PSUs also have fans that don't spin at top speed
>> the whole time, given that they sound like jet engines when they do.
>
> Well, if the thermal speed control built into a cheap Octigen ATX PSU
> I cannibalised recently to repair the PSU in my desktop machine is
> anything to go by, then I'm not surprised they suffer short lives from
> overheating. In this case, the slow speed setting was far too slow and
> the thermal cut in point was set so high the top of the tower became
> very warm to the touch (causing the CPU fan to resort to full speed to
> compensate for the very high temps never ever seen before with my
> previous PSU setup).
I'd expect the top of the tower to become warm: so does mine. The
tower's made of metal and has a large surface exposed to the outside
air: dumping heat into it is a fairly efficient way of getting rid of
it.
The CPU temperature never goes above 70C even under the most extreme
load, and idles just below 50C. That's nowhere near the point of
failure. Hell, my old Athlon 4 got hotter than that.
The disk temperatures hang around 25C--30C, which is if anything too
*low* (Google's large-scale analysis suggests that a drive temp of
35C--40C is correlated with maximum lifespan).
> The PSU itself was almost too hot to touch
OK, that's definitely not enough cooling then. :)
> Regarding the 145W unit in the Gateway 2000 case, I've never heard it
> speed up (except possibly during heatwave conditions a few years back
> when the room temperature got above the 27 dec C mark) and this very
> nicely keeps all four drives below the 13 deg above ambient mark.
Yeah, I can't really tell which fans speed up when. Mostly it's the CPU
fans that are temperature-dependent, but a couple of times, after hours
of maxing out all cores during a heatwave, I've heard the disk fan and
then the PSU fan on the RAID server spin up, and it's an unmistakable
howl. (Interestingly, unlike the CPU fan, these do *not* start at
maximum speed at bootup.)
>> I see no behaviour problems with my WD10EACS. What misbehaviour do you
>> see? Does it actually stop sending you data and spin all the way down
>> after only 8.3s? That seems very unlikely: are you sure you didn't
>> misinterpret the spindown-to-2k-RPM setting as a complete spindown?
>
> No, not spindown. The issue isn't the very slight response delay so
> much as the 300,000 head unload life limit rating and the fact that,
> with the unconscionable default 8 second time-out, this can be reached
> in as little as 6 months or less of use.
Interesting. The heads on my drives don't unload at all. Here's some
SMART data with a couple of columns trimmed out:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 0
3 Spin_Up_Time 0x0027 163 163 021 6808
4 Start_Stop_Count 0x0032 100 100 000 88
5 Reallocated_Sector_Ct 0x0033 200 200 140 0
7 Seek_Error_Rate 0x002e 100 253 051 0
9 Power_On_Hours 0x0032 072 072 000 20457
10 Spin_Retry_Count 0x0032 100 253 051 0
11 Calibration_Retry_Count 0x0032 100 253 051 0
12 Power_Cycle_Count 0x0032 100 100 000 88
192 Power-Off_Retract_Count 0x0032 200 200 000 8
193 Load_Cycle_Count 0x0032 200 200 000 88
194 Temperature_Celsius 0x0022 120 113 000 30
196 Reallocated_Event_Count 0x0032 200 200 000 0
197 Current_Pending_Sector 0x0032 200 200 000 0
198 Offline_Uncorrectable 0x0030 200 200 000 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 0
200 Multi_Zone_Error_Rate 0x0008 200 200 051 0
(This is assuming that the Areca controller is not lying to me about the
disks it fronts, and I see no reason why it should.)
So I see one load per power-cycle, which is perfectly OK.
> As a matter of fact (and I can't work out why, considering the load
> cycle count of 9572 on the second unit, commissioned only just 85
> hours later) the first of the WD20EARS drives now shows a load cycle
> count figure of 366542 (up from the 366539 figure noted just 21 hours
OW. If it really is unloading the heads that often, I'd call that bloody
horrible. My drives are reported as
=== START OF INFORMATION SECTION ===
Model Family: Western Digital Caviar Green family
Device Model: WDC WD10EACS-65D6B0
Firmware Version: 01.01A01
What are yours?
> If the head unload rating life rating was 3,000,000, it wouldn't be
> an issue but, even taking WD's revised estimate of 600,000 into
> account, this still represents a possible failure within a mere 12
> months of use so it's vital that the time-out be raised to something
> more reasonable (like 300 seconds!) if you want to give the drive a
> fighting chance to survive the 3 (5?) year warranty period.
I'm surprised they're resorting to head unloading at all, given that
these drives are meant to respond immediately to requests even in
low-power mode. Mine (bought in 2009) don't unload. (But maybe the
controller needs to tell the drive not to do that, and mine does that
automatically?)
> However, head unloading time out issues aside, it's the strange non
> eco-green power consumption curve that irks me the most. When I
> installed those drives, I used them to replace a couple of 7200 rpm
> 1TB Samsung spinpoints that lived alongside a couple of 1TB Samsung
> 5400 rpm eco green drives, expecting the 53 watt consumption to fall
> below the 49 watt mark but all that happened was that the power varied
> between 48 and 53 watts (infrequently as low as 48, most often 53
> watts regardless of how long the box was effectively 'idle').
Now that I'd expect from a drive that works like mine. The power
consumption is reduced when the drives are spun down to a reduced RPM,
at the cost of increased latency and briefly increased power consumption
while spinning up (much of which is got back when the drive spins down
again, so you get a dip in power consumption below even the idling
figure at that point).
> As it happened, the controller PCB transplant trick worked just fine
> and I successfully retrieved all the data. The problem was that I
> under-estimated the PCB's susceptibility to ESD[2] when it came to
> refitting it back to the original drive it had been borrowed from and
> I now have two dead Samsung drives with perfect mechanicals for my
> trouble.
See, this is why I never touch hardware (well, other than the
coordination deficits). I seem to be entirely made of static
electricity: I have blown things simply touching machines' cases...
> [1] What I'd initially assumed was a 4k sector issue with win2k (even
> after correctly aligning the partition boundary) looks like it may
> have had more to do with an e-SATA issue within the win2k box since I
> saw exactly the same slow and stuttered write performance along with
> mouse movement hangs when I tested the newly released and
> re-partitioned 512 byte sectored 2TB eco-green Samsung drive in an
> e-SATA enclosure.
Mouse movement hangs?! If you weren't swapping really heavily, that
suggests it was saturating or locking up the bus for a long period of
time. Gah.
--
NULL && (void)
> On Fri, 21 Oct 2011 15:33:15 +0100, Nix <nix-ra...@esperi.org.uk> wrote:
>
>> On 21 Oct 2011, Johny B. Good outgrape:
>>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>>> takes between 46 and 52 watts when idle (no spin down power saving
>>> involved) depending on what 'mood' the two 'so called eco green'
>>> Western Digital WD20EARS drives[2] happen to be in.
>
> First off, sorry for the late response. Secondly, oops! When I used
> the phrase "jbod, not RAID", I was mis-using it as shorthand for
> separate disk volumes on the server box rather than a single gigantic
> 8(short)TB disk volume.
>> at low. My previous aged PCI software RAID array could manage 30Mb/s if
>> you were very lucky...
>
> Oh, how things have changed!
that, and the multiple layers of converters hanging off it to allow it
to connect to anything (welcome to SCSI hell).
>>> [1] You might think that an old small form factor 145W rated ATX PSU
>>> might succumb to overload on its 4.2A rated 12v rail with all those
>>> hard disk drives but it takes the spin up loading in its stride (even
>>
>> Yeah, four disks are unlikely to cause problems. Eight might though. (My
>> RAID array is a hardware one and has an option to stagger spinup
>> specifically to avoid overloading the PSU.)
>
> That's a given, especially true of SCSI arrays.
the drives and to the system as a whole, though obviously normal AHCI
controllers don't have a gig of cache and their own fan, web server, and
SNMP support).
>> One would hope newer PSUs also have fans that don't spin at top speed
>> the whole time, given that they sound like jet engines when they do.
>
> Well, if the thermal speed control built into a cheap Octigen ATX PSU
> I cannibalised recently to repair the PSU in my desktop machine is
> anything to go by, then I'm not surprised they suffer short lives from
> overheating. In this case, the slow speed setting was far too slow and
> the thermal cut in point was set so high the top of the tower became
> very warm to the touch (causing the CPU fan to resort to full speed to
> compensate for the very high temps never ever seen before with my
> previous PSU setup).
tower's made of metal and has a large surface exposed to the outside
air: dumping heat into it is a fairly efficient way of getting rid of
it.
The CPU temperature never goes above 70C even under the most extreme
load, and idles just below 50C. That's nowhere near the point of
failure. Hell, my old Athlon 4 got hotter than that.
The disk temperatures hang around 25C--30C, which is if anything too
*low* (Google's large-scale analysis suggests that a drive temp of
35C--40C is correlated with maximum lifespan).
> The PSU itself was almost too hot to touch
> Regarding the 145W unit in the Gateway 2000 case, I've never heard it
> speed up (except possibly during heatwave conditions a few years back
> when the room temperature got above the 27 dec C mark) and this very
> nicely keeps all four drives below the 13 deg above ambient mark.
fans that are temperature-dependent, but a couple of times, after hours
of maxing out all cores during a heatwave, I've heard the disk fan and
then the PSU fan on the RAID server spin up, and it's an unmistakable
howl. (Interestingly, unlike the CPU fan, these do *not* start at
maximum speed at bootup.)
>> I see no behaviour problems with my WD10EACS. What misbehaviour do you
>> see? Does it actually stop sending you data and spin all the way down
>> after only 8.3s? That seems very unlikely: are you sure you didn't
>> misinterpret the spindown-to-2k-RPM setting as a complete spindown?
>
> No, not spindown. The issue isn't the very slight response delay so
> much as the 300,000 head unload life limit rating and the fact that,
> with the unconscionable default 8 second time-out, this can be reached
> in as little as 6 months or less of use.
SMART data with a couple of columns trimmed out:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 0
3 Spin_Up_Time 0x0027 163 163 021 6808
4 Start_Stop_Count 0x0032 100 100 000 88
5 Reallocated_Sector_Ct 0x0033 200 200 140 0
7 Seek_Error_Rate 0x002e 100 253 051 0
9 Power_On_Hours 0x0032 072 072 000 20457
10 Spin_Retry_Count 0x0032 100 253 051 0
11 Calibration_Retry_Count 0x0032 100 253 051 0
12 Power_Cycle_Count 0x0032 100 100 000 88
192 Power-Off_Retract_Count 0x0032 200 200 000 8
193 Load_Cycle_Count 0x0032 200 200 000 88
194 Temperature_Celsius 0x0022 120 113 000 30
196 Reallocated_Event_Count 0x0032 200 200 000 0
197 Current_Pending_Sector 0x0032 200 200 000 0
198 Offline_Uncorrectable 0x0030 200 200 000 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 0
200 Multi_Zone_Error_Rate 0x0008 200 200 051 0
(This is assuming that the Areca controller is not lying to me about the
disks it fronts, and I see no reason why it should.)
So I see one load per power-cycle, which is perfectly OK.
> As a matter of fact (and I can't work out why, considering the load
> cycle count of 9572 on the second unit, commissioned only just 85
> hours later) the first of the WD20EARS drives now shows a load cycle
> count figure of 366542 (up from the 366539 figure noted just 21 hours
horrible. My drives are reported as
=== START OF INFORMATION SECTION ===
Model Family: Western Digital Caviar Green family
Device Model: WDC WD10EACS-65D6B0
Firmware Version: 01.01A01
What are yours?
> If the head unload rating life rating was 3,000,000, it wouldn't be
> an issue but, even taking WD's revised estimate of 600,000 into
> account, this still represents a possible failure within a mere 12
> months of use so it's vital that the time-out be raised to something
> more reasonable (like 300 seconds!) if you want to give the drive a
> fighting chance to survive the 3 (5?) year warranty period.
these drives are meant to respond immediately to requests even in
low-power mode. Mine (bought in 2009) don't unload. (But maybe the
controller needs to tell the drive not to do that, and mine does that
automatically?)
> However, head unloading time out issues aside, it's the strange non
> eco-green power consumption curve that irks me the most. When I
> installed those drives, I used them to replace a couple of 7200 rpm
> 1TB Samsung spinpoints that lived alongside a couple of 1TB Samsung
> 5400 rpm eco green drives, expecting the 53 watt consumption to fall
> below the 49 watt mark but all that happened was that the power varied
> between 48 and 53 watts (infrequently as low as 48, most often 53
> watts regardless of how long the box was effectively 'idle').
consumption is reduced when the drives are spun down to a reduced RPM,
at the cost of increased latency and briefly increased power consumption
while spinning up (much of which is got back when the drive spins down
again, so you get a dip in power consumption below even the idling
figure at that point).
> As it happened, the controller PCB transplant trick worked just fine
> and I successfully retrieved all the data. The problem was that I
> under-estimated the PCB's susceptibility to ESD[2] when it came to
> refitting it back to the original drive it had been borrowed from and
> I now have two dead Samsung drives with perfect mechanicals for my
> trouble.
coordination deficits). I seem to be entirely made of static
electricity: I have blown things simply touching machines' cases...
> [1] What I'd initially assumed was a 4k sector issue with win2k (even
> after correctly aligning the partition boundary) looks like it may
> have had more to do with an e-SATA issue within the win2k box since I
> saw exactly the same slow and stuttered write performance along with
> mouse movement hangs when I tested the newly released and
> re-partitioned 512 byte sectored 2TB eco-green Samsung drive in an
> e-SATA enclosure.
suggests it was saturating or locking up the bus for a long period of
time. Gah.
--
NULL && (void)
Nix
Oct 28, 2011, 10:52:14 AM10/28/11
to
On 28 Oct 2011, Nix spake thusly:
controller has an option to turn spindown-on-inactive off or on. (Off
by default, naturally.)
Does your head unloading correlate with a spindown, or is it just
unloading on its own?
--
NULL && (void)
> I'm surprised they're resorting to head unloading at all, given that
> these drives are meant to respond immediately to requests even in
> low-power mode. Mine (bought in 2009) don't unload. (But maybe the
> controller needs to tell the drive not to do that, and mine does that
> automatically?)
This may well be at least partially related to the controller: my Areca
> these drives are meant to respond immediately to requests even in
> low-power mode. Mine (bought in 2009) don't unload. (But maybe the
> controller needs to tell the drive not to do that, and mine does that
> automatically?)
controller has an option to turn spindown-on-inactive off or on. (Off
by default, naturally.)
Does your head unloading correlate with a spindown, or is it just
unloading on its own?
--
NULL && (void)
Johny B Good
Oct 28, 2011, 3:34:34 PM10/28/11
to
On Fri, 28 Oct 2011 15:31:03 +0100, Nix <nix-ra...@esperi.org.uk>
from several drives. I just wanted to clarify _my_ usage of the expression
'JBOD' to avoid the confusion that seems to abound with the meaning of
this acronym.
>
>>> at low. My previous aged PCI software RAID array could manage 30Mb/s if
>>> you were very lucky...
>>
>> Oh, how things have changed!
>
> I think we can probably blame the sym53c875 component of the array for
> that, and the multiple layers of converters hanging off it to allow it
> to connect to anything (welcome to SCSI hell).
>
>>>> [1] You might think that an old small form factor 145W rated ATX PSU
>>>> might succumb to overload on its 4.2A rated 12v rail with all those
>>>> hard disk drives but it takes the spin up loading in its stride (even
>>>
>>> Yeah, four disks are unlikely to cause problems. Eight might though.
>>> (My
>>> RAID array is a hardware one and has an option to stagger spinup
>>> specifically to avoid overloading the PSU.)
>>
>> That's a given, especially true of SCSI arrays.
>
> Mine's not SCSI, just normal AHCI (well, it *appears* as normal AHCI to
> the drives and to the system as a whole, though obviously normal AHCI
> controllers don't have a gig of cache and their own fan, web server, and
> SNMP support).
Well, device manager in win2k lumps the AHCI driver as a "SCSI and RAID
Controllers" device.
>
>>> One would hope newer PSUs also have fans that don't spin at top speed
>>> the whole time, given that they sound like jet engines when they do.
>>
>> Well, if the thermal speed control built into a cheap Octigen ATX PSU
>> I cannibalised recently to repair the PSU in my desktop machine is
>> anything to go by, then I'm not surprised they suffer short lives from
>> overheating. In this case, the slow speed setting was far too slow and
>> the thermal cut in point was set so high the top of the tower became
>> very warm to the touch (causing the CPU fan to resort to full speed to
>> compensate for the very high temps never ever seen before with my
>> previous PSU setup).
>
> I'd expect the top of the tower to become warm: so does mine. The
> tower's made of metal and has a large surface exposed to the outside
> air: dumping heat into it is a fairly efficient way of getting rid of
> it.
Not for a typical PC I'm afraid, that adds a second temperature gradient
between the innards of the box and its outside environment. Using a flow
of air between the inside and the outside of the PC case can eliminate
this extra gradient when done properly.
If the components are rated for the much higher temperatures within a
sealed box that relies on passive cooling alone to dissipate heat
transferred to the outer surface of its enclosure, then all is well and
good.
Unfortunately, most PC designs create too high a heat load for this
method alone to maintain a sufficiently low enough temperature to allow
the temperature sensitive components (in particular, hard disk drives) to
stay below their maximum rated temperature limit and it has been normal
practice since the very first PC to use ventilated cases (typically
ventilated by the PSU fan alone in designs with idle consumption power
levels below the hundred watt mark).
Checking with an IR thermometer, the top of the tower case is only about
6 or 7 degrees above ambient, barely warmer than the sides of the case
(which feels cool) when checked by touch. This is as it should be. A case
that feels warm to the touch is likely to be a good 15 to 20 degrees above
ambient, implying a similar temperature gradient on the inside of the box
which could mean internal air temperatures around the 50 deg mark,
assuming a room temp of 20 deg. This leaves very little margin for things
like disk drives, even assuming they're sitting in a cooler part of the
box.
>
> The CPU temperature never goes above 70C even under the most extreme
> load, and idles just below 50C. That's nowhere near the point of
> failure. Hell, my old Athlon 4 got hotter than that.
That's not an issue, unless you're looking at service lifetimes in excess
of ten years. 90 deg C is just fine for a CPU provided it isn't marginal
on clock speed (rated or over clocked) where the higher on resistances of
the cmos gates will reduce switching speeds.
The one thing that does limit internal temperature is the hard disk drive
which is likely to overheat if the internal air temperature gets as high
as 50 deg C. That's a very good reason right there for fan cooling the
innards of a PC whether or not a liquid cooling solution is used for the
CPU/GPU/Chipset components (you still need to get rid of the heat produced
by disk drives and the other components not specifically covered by the
liquid cooling).
>
> The disk temperatures hang around 25C--30C, which is if anything too
> *low* (Google's large-scale analysis suggests that a drive temp of
> 35C--40C is correlated with maximum lifespan).
AFAIR, I don't think this was a very pronounced effect and it's worth
noting that this was in the context of always spinning 24/7 operation.
whilst this is more relevant to a home server, that's certainly not true
of a desktop machine which may only run for 8 to 12 hours a day going
through one or more temperature cycles per day where the lower temperature
will be more beneficial to reducing the effect of thermal cycling induced
stress.
Ok, I'll bite! ;-)
=========================================================================================================
Device /dev/ad4 - Hitachi HDS5C3030ALA630/MEAOA5C0
Device Model: Hitachi HDS5C3030ALA630
Serial Number: *************
LU WWN Device Id: 5 000cca 228c1cae2
Firmware Version: MEAOA5C0
User Capacity: 3,000,592,982,016 bytes [3.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: ATA-8-ACS revision 4
Local Time is: Fri Oct 28 18:11:19 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x82) Offline data collection activity
was completed without error.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (36667) seconds.
Offline data collection
capabilities: (0x5b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
No Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 1) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
SCT capabilities: (0x003d) SCT Status supported.
SCT Error Recovery Control supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x000b 100 100 016 Pre-fail
Always - 0
2 Throughput_Performance 0x0005 132 132 054 Pre-fail
Offline - 117
3 Spin_Up_Time 0x0007 100 100 024 Pre-fail
Always - 414
4 Start_Stop_Count 0x0012 100 100 000 Old_age
Always - 8
5 Reallocated_Sector_Ct 0x0033 100 100 005 Pre-fail
Always - 0
7 Seek_Error_Rate 0x000b 100 100 067 Pre-fail
Always - 0
8 Seek_Time_Performance 0x0005 135 135 020 Pre-fail
Offline - 31
9 Power_On_Hours 0x0012 100 100 000 Old_age
Always - 70
10 Spin_Retry_Count 0x0013 100 100 060 Pre-fail
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 6
192 Power-Off_Retract_Count 0x0032 100 100 000 Old_age
Always - 38
193 Load_Cycle_Count 0x0012 100 100 000 Old_age
Always - 38
194 Temperature_Celsius 0x0002 200 200 000 Old_age
Always - 30 (Min/Max 22/37)
196 Reallocated_Event_Count 0x0032 100 100 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0022 100 100 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0008 100 100 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x000a 200 200 000 Old_age
Always - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad6 - SAMSUNG HD203WI/1AN10003
Device Model: SAMSUNG HD203WI
Serial Number: *************
LU WWN Device Id: 5 0024e9 003d1002d
Firmware Version: 1AN10003
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: ATA-8-ACS revision 6
Local Time is: Fri Oct 28 18:11:21 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x00) Offline data collection activity
was never started.
Auto Offline Data Collection: Disabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (25500) seconds.
Offline data collection
capabilities: (0x5b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
No Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
SCT capabilities: (0x003f) SCT Status supported.
SCT Error Recovery Control supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 100 100 051 Pre-fail
Always - 2
2 Throughput_Performance 0x0026 252 252 000 Old_age
Always - 0
3 Spin_Up_Time 0x0023 062 061 025 Pre-fail
Always - 11770
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 32
5 Reallocated_Sector_Ct 0x0033 252 252 010 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 252 252 051 Old_age
Always - 0
8 Seek_Time_Performance 0x0024 252 252 015 Old_age
Offline - 0
9 Power_On_Hours 0x0032 100 100 000 Old_age
Always - 9068
10 Spin_Retry_Count 0x0032 252 252 051 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 252 252 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 27
191 G-Sense_Error_Rate 0x0022 100 100 000 Old_age
Always - 5
192 Power-Off_Retract_Count 0x0022 252 252 000 Old_age
Always - 0
194 Temperature_Celsius 0x0002 064 062 000 Old_age
Always - 27 (Min/Max 20/39)
195 Hardware_ECC_Recovered 0x003a 100 100 000 Old_age
Always - 0
196 Reallocated_Event_Count 0x0032 252 252 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 252 252 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 252 252 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0036 200 200 000 Old_age
Always - 0
200 Multi_Zone_Error_Rate 0x002a 100 100 000 Old_age
Always - 0
223 Load_Retry_Count 0x0032 252 252 000 Old_age
Always - 0
225 Load_Cycle_Count 0x0032 097 097 000 Old_age
Always - 40377
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
Note: selective self-test log revision number (0) not 1 implies that no
selective self-test has ever been run
SMART Selective self-test log data structure revision number 0
Note: revision number not 1 implies that no selective self-test has ever
been run
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Completed [00% left] (0-65535)
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad8 - WDC WD20EARS-00S8B1/80.00A80
Device Model: WDC WD20EARS-00S8B1
Serial Number: ******************
LU WWN Device Id: 5 0014ee 2595827c0
Firmware Version: 80.00A80
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Fri Oct 28 18:11:23 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x84) Offline data collection activity
was suspended by an interrupting command from host.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (43200) seconds.
Offline data collection
capabilities: (0x7b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
Conveyance self-test routine
recommended polling time: ( 5) minutes.
SCT capabilities: (0x3031) SCT Status supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail
Always - 0
3 Spin_Up_Time 0x0027 152 141 021 Pre-fail
Always - 9391
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 562
5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 200 200 000 Old_age
Always - 0
9 Power_On_Hours 0x0032 082 082 000 Old_age
Always - 13428
10 Spin_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 205
192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age
Always - 154
193 Load_Cycle_Count 0x0032 197 197 000 Old_age
Always - 9707
194 Temperature_Celsius 0x0022 120 108 000 Old_age
Always - 32
196 Reallocated_Event_Count 0x0032 200 200 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 200 200 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 200 200 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age
Always - 7
200 Multi_Zone_Error_Rate 0x0008 200 200 000 Old_age
Offline - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad10 - WDC WD20EARS-00S8B1/80.00A80
Device Model: WDC WD20EARS-00S8B1
Serial Number: ***************
LU WWN Device Id: 5 0014ee 2aeadc72d
Firmware Version: 80.00A80
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Fri Oct 28 18:11:24 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x82) Offline data collection activity
was completed without error.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (42360) seconds.
Offline data collection
capabilities: (0x7b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
Conveyance self-test routine
recommended polling time: ( 5) minutes.
SCT capabilities: (0x3031) SCT Status supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail
Always - 0
3 Spin_Up_Time 0x0027 157 144 021 Pre-fail
Always - 9133
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 529
5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 200 200 000 Old_age
Always - 0
9 Power_On_Hours 0x0032 082 082 000 Old_age
Always - 13513
10 Spin_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 199
192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age
Always - 158
193 Load_Cycle_Count 0x0032 078 078 000 Old_age
Always - 366543
194 Temperature_Celsius 0x0022 123 107 000 Old_age
Always - 29
196 Reallocated_Event_Count 0x0032 200 200 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 200 200 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 200 200 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age
Always - 9
200 Multi_Zone_Error_Rate 0x0008 200 200 000 Old_age
Offline - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
==========================================================================================================
That's the complete listing for all four spinning disk drives.
I've blanked the serial numbers and excluded the card reader slots.
>
>> If the head unload rating life rating was 3,000,000, it wouldn't be
>> an issue but, even taking WD's revised estimate of 600,000 into
>> account, this still represents a possible failure within a mere 12
>> months of use so it's vital that the time-out be raised to something
>> more reasonable (like 300 seconds!) if you want to give the drive a
>> fighting chance to survive the 3 (5?) year warranty period.
>
> I'm surprised they're resorting to head unloading at all, given that
> these drives are meant to respond immediately to requests even in
> low-power mode. Mine (bought in 2009) don't unload. (But maybe the
> controller needs to tell the drive not to do that, and mine does that
> automatically?)
It was one of the power saving features. The heads are unloaded (but not
latched in the parked state) so that the track servo controller can be
disabled. Why the hell they chose a time out interval of only 8 seconds,
God alone knows. It should have been quite obvious that the 300,000 unload
cycles limit was likely to be reached within less than 6 months of
operation. Just doubling the interval would most likely have extended the
operational hours per 300,000 unload cycles by a factor of three or more.
Better still if they had chosen a much larger time out default figure of
30 seconds or more, leaving the end user to decide their own 'optimum'
time out value. Hopefully, the newer 3TB drives have this feature disabled
by default (along with a more eco-friendly power consumption regime).
>
>> However, head unloading time out issues aside, it's the strange non
>> eco-green power consumption curve that irks me the most. When I
>> installed those drives, I used them to replace a couple of 7200 rpm
>> 1TB Samsung spinpoints that lived alongside a couple of 1TB Samsung
>> 5400 rpm eco green drives, expecting the 53 watt consumption to fall
>> below the 49 watt mark but all that happened was that the power varied
>> between 48 and 53 watts (infrequently as low as 48, most often 53
>> watts regardless of how long the box was effectively 'idle').
>
> Now that I'd expect from a drive that works like mine. The power
> consumption is reduced when the drives are spun down to a reduced RPM,
> at the cost of increased latency and briefly increased power consumption
> while spinning up (much of which is got back when the drive spins down
> again, so you get a dip in power consumption below even the idling
> figure at that point).
I've never been attracted by the spin down power saving feature,
preferring the instant availability of data and, most importantly, the
improved reliability of constant spin (minimal temperature variations).
>
>> As it happened, the controller PCB transplant trick worked just fine
>> and I successfully retrieved all the data. The problem was that I
>> under-estimated the PCB's susceptibility to ESD[2] when it came to
>> refitting it back to the original drive it had been borrowed from and
>> I now have two dead Samsung drives with perfect mechanicals for my
>> trouble.
>
> See, this is why I never touch hardware (well, other than the
> coordination deficits). I seem to be entirely made of static
> electricity: I have blown things simply touching machines' cases...
I haven't 'zapped' any kit for many years due to an ESD handling
awareness that I developed over thirty years ago. Even now, I'm only
assuming the problem was an ESD accident since it seems rather extreme of
Samsung to have incorporate an "Only One Byte of The Cherry" limit on
controller board transplantation operations (but, who knows? Anything is
possible given microprocessor control and some flash ram).
>
>> [1] What I'd initially assumed was a 4k sector issue with win2k (even
>> after correctly aligning the partition boundary) looks like it may
>> have had more to do with an e-SATA issue within the win2k box since I
>> saw exactly the same slow and stuttered write performance along with
>> mouse movement hangs when I tested the newly released and
>> re-partitioned 512 byte sectored 2TB eco-green Samsung drive in an
>> e-SATA enclosure.
>
> Mouse movement hangs?! If you weren't swapping really heavily, that
> suggests it was saturating or locking up the bus for a long period of
> time. Gah.
Gah indeed! That told me something was _very_ wrong but, when I first saw
these symptoms, I put two and two together and got Five (must have been
large values of two), assuming it was something to do with it being an
effect of the 4k AF used by the WD20EARS drive I was testing. It hadn't
occurred to me that I could have been witnessing the effects of bad SATA
connector flakiness otherwise I would have transplanted it straight into
the desktop machine and bought a third 2TB Samsung eco green drive to take
its place in the server box.
As things are, I won't get the chance to verify whether or not the 4k AF
is an issue under win2k for some time, now that I've splashed out on that
5K3000 Hitachi drive. Between the extra 1TB added to the server box and
the 1TB upgrade on the desktop machine, I should have enough capacity to
keep me going for another 6 to 8 months, by which time (hopefully), the
current price hike on disk drives should be just another bad memory.
--
Regards JB Good
wrote:
> On 28 Oct 2011, Johny B. Good stated:
>
>> On Fri, 21 Oct 2011 15:33:15 +0100, Nix <nix-ra...@esperi.org.uk>
>> wrote:
>>
> On 28 Oct 2011, Johny B. Good stated:
>
>> On Fri, 21 Oct 2011 15:33:15 +0100, Nix <nix-ra...@esperi.org.uk>
>> wrote:
>>
>>> On 21 Oct 2011, Johny B. Good wrote:
>>>>
>>>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>>>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>>>> takes between 46 and 52 watts when idle (no spin down power saving
>>>> involved) depending on what 'mood' the two 'so called eco green'
>>>> Western Digital WD20EARS drives[2] happen to be in.
>>
>> First off, sorry for the late response. Secondly, oops! When I used
>> the phrase "jbod, not RAID", I was mis-using it as shorthand for
>> separate disk volumes on the server box rather than a single gigantic
>> 8(short)TB disk volume.
>
> That *is* JBOD, 'just a bunch of disks', not even RAID-0.
I've seen the use of 'JBOD' RAID to describe a single disk volume created
>>>>
>>>> I'm using an ancient Gateway 2000 desktop case with its original 145W
>>>> ATX PSU[1] to house my 4 drive (jbod, not RAID) FreeNAS server. This
>>>> takes between 46 and 52 watts when idle (no spin down power saving
>>>> involved) depending on what 'mood' the two 'so called eco green'
>>>> Western Digital WD20EARS drives[2] happen to be in.
>>
>> First off, sorry for the late response. Secondly, oops! When I used
>> the phrase "jbod, not RAID", I was mis-using it as shorthand for
>> separate disk volumes on the server box rather than a single gigantic
>> 8(short)TB disk volume.
>
> That *is* JBOD, 'just a bunch of disks', not even RAID-0.
from several drives. I just wanted to clarify _my_ usage of the expression
'JBOD' to avoid the confusion that seems to abound with the meaning of
this acronym.
>
>>> at low. My previous aged PCI software RAID array could manage 30Mb/s if
>>> you were very lucky...
>>
>> Oh, how things have changed!
>
> I think we can probably blame the sym53c875 component of the array for
> that, and the multiple layers of converters hanging off it to allow it
> to connect to anything (welcome to SCSI hell).
>
>>>> [1] You might think that an old small form factor 145W rated ATX PSU
>>>> might succumb to overload on its 4.2A rated 12v rail with all those
>>>> hard disk drives but it takes the spin up loading in its stride (even
>>>
>>> Yeah, four disks are unlikely to cause problems. Eight might though.
>>> (My
>>> RAID array is a hardware one and has an option to stagger spinup
>>> specifically to avoid overloading the PSU.)
>>
>> That's a given, especially true of SCSI arrays.
>
> Mine's not SCSI, just normal AHCI (well, it *appears* as normal AHCI to
> the drives and to the system as a whole, though obviously normal AHCI
> controllers don't have a gig of cache and their own fan, web server, and
> SNMP support).
Controllers" device.
>
>>> One would hope newer PSUs also have fans that don't spin at top speed
>>> the whole time, given that they sound like jet engines when they do.
>>
>> Well, if the thermal speed control built into a cheap Octigen ATX PSU
>> I cannibalised recently to repair the PSU in my desktop machine is
>> anything to go by, then I'm not surprised they suffer short lives from
>> overheating. In this case, the slow speed setting was far too slow and
>> the thermal cut in point was set so high the top of the tower became
>> very warm to the touch (causing the CPU fan to resort to full speed to
>> compensate for the very high temps never ever seen before with my
>> previous PSU setup).
>
> I'd expect the top of the tower to become warm: so does mine. The
> tower's made of metal and has a large surface exposed to the outside
> air: dumping heat into it is a fairly efficient way of getting rid of
> it.
between the innards of the box and its outside environment. Using a flow
of air between the inside and the outside of the PC case can eliminate
this extra gradient when done properly.
If the components are rated for the much higher temperatures within a
sealed box that relies on passive cooling alone to dissipate heat
transferred to the outer surface of its enclosure, then all is well and
good.
Unfortunately, most PC designs create too high a heat load for this
method alone to maintain a sufficiently low enough temperature to allow
the temperature sensitive components (in particular, hard disk drives) to
stay below their maximum rated temperature limit and it has been normal
practice since the very first PC to use ventilated cases (typically
ventilated by the PSU fan alone in designs with idle consumption power
levels below the hundred watt mark).
Checking with an IR thermometer, the top of the tower case is only about
6 or 7 degrees above ambient, barely warmer than the sides of the case
(which feels cool) when checked by touch. This is as it should be. A case
that feels warm to the touch is likely to be a good 15 to 20 degrees above
ambient, implying a similar temperature gradient on the inside of the box
which could mean internal air temperatures around the 50 deg mark,
assuming a room temp of 20 deg. This leaves very little margin for things
like disk drives, even assuming they're sitting in a cooler part of the
box.
>
> The CPU temperature never goes above 70C even under the most extreme
> load, and idles just below 50C. That's nowhere near the point of
> failure. Hell, my old Athlon 4 got hotter than that.
of ten years. 90 deg C is just fine for a CPU provided it isn't marginal
on clock speed (rated or over clocked) where the higher on resistances of
the cmos gates will reduce switching speeds.
The one thing that does limit internal temperature is the hard disk drive
which is likely to overheat if the internal air temperature gets as high
as 50 deg C. That's a very good reason right there for fan cooling the
innards of a PC whether or not a liquid cooling solution is used for the
CPU/GPU/Chipset components (you still need to get rid of the heat produced
by disk drives and the other components not specifically covered by the
liquid cooling).
>
> The disk temperatures hang around 25C--30C, which is if anything too
> *low* (Google's large-scale analysis suggests that a drive temp of
> 35C--40C is correlated with maximum lifespan).
noting that this was in the context of always spinning 24/7 operation.
whilst this is more relevant to a home server, that's certainly not true
of a desktop machine which may only run for 8 to 12 hours a day going
through one or more temperature cycles per day where the lower temperature
will be more beneficial to reducing the effect of thermal cycling induced
stress.
=========================================================================================================
Device /dev/ad4 - Hitachi HDS5C3030ALA630/MEAOA5C0
=== START OF INFORMATION SECTION ===
Model Family: Hitachi Deskstar 5K3000
Device Model: Hitachi HDS5C3030ALA630
Serial Number: *************
LU WWN Device Id: 5 000cca 228c1cae2
Firmware Version: MEAOA5C0
User Capacity: 3,000,592,982,016 bytes [3.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: ATA-8-ACS revision 4
Local Time is: Fri Oct 28 18:11:19 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x82) Offline data collection activity
was completed without error.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (36667) seconds.
Offline data collection
capabilities: (0x5b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
No Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 1) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
SCT capabilities: (0x003d) SCT Status supported.
SCT Error Recovery Control supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x000b 100 100 016 Pre-fail
Always - 0
2 Throughput_Performance 0x0005 132 132 054 Pre-fail
Offline - 117
3 Spin_Up_Time 0x0007 100 100 024 Pre-fail
Always - 414
4 Start_Stop_Count 0x0012 100 100 000 Old_age
Always - 8
5 Reallocated_Sector_Ct 0x0033 100 100 005 Pre-fail
Always - 0
7 Seek_Error_Rate 0x000b 100 100 067 Pre-fail
Always - 0
8 Seek_Time_Performance 0x0005 135 135 020 Pre-fail
Offline - 31
9 Power_On_Hours 0x0012 100 100 000 Old_age
Always - 70
10 Spin_Retry_Count 0x0013 100 100 060 Pre-fail
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 6
192 Power-Off_Retract_Count 0x0032 100 100 000 Old_age
Always - 38
193 Load_Cycle_Count 0x0012 100 100 000 Old_age
Always - 38
194 Temperature_Celsius 0x0002 200 200 000 Old_age
Always - 30 (Min/Max 22/37)
196 Reallocated_Event_Count 0x0032 100 100 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0022 100 100 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0008 100 100 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x000a 200 200 000 Old_age
Always - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad6 - SAMSUNG HD203WI/1AN10003
=== START OF INFORMATION SECTION ===
Model Family: SAMSUNG SpinPoint F3 EG
Device Model: SAMSUNG HD203WI
Serial Number: *************
LU WWN Device Id: 5 0024e9 003d1002d
Firmware Version: 1AN10003
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: ATA-8-ACS revision 6
Local Time is: Fri Oct 28 18:11:21 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x00) Offline data collection activity
was never started.
Auto Offline Data Collection: Disabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (25500) seconds.
Offline data collection
capabilities: (0x5b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
No Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
SCT capabilities: (0x003f) SCT Status supported.
SCT Error Recovery Control supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 100 100 051 Pre-fail
Always - 2
2 Throughput_Performance 0x0026 252 252 000 Old_age
Always - 0
3 Spin_Up_Time 0x0023 062 061 025 Pre-fail
Always - 11770
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 32
5 Reallocated_Sector_Ct 0x0033 252 252 010 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 252 252 051 Old_age
Always - 0
8 Seek_Time_Performance 0x0024 252 252 015 Old_age
Offline - 0
9 Power_On_Hours 0x0032 100 100 000 Old_age
Always - 9068
10 Spin_Retry_Count 0x0032 252 252 051 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 252 252 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 27
191 G-Sense_Error_Rate 0x0022 100 100 000 Old_age
Always - 5
192 Power-Off_Retract_Count 0x0022 252 252 000 Old_age
Always - 0
194 Temperature_Celsius 0x0002 064 062 000 Old_age
Always - 27 (Min/Max 20/39)
195 Hardware_ECC_Recovered 0x003a 100 100 000 Old_age
Always - 0
196 Reallocated_Event_Count 0x0032 252 252 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 252 252 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 252 252 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0036 200 200 000 Old_age
Always - 0
200 Multi_Zone_Error_Rate 0x002a 100 100 000 Old_age
Always - 0
223 Load_Retry_Count 0x0032 252 252 000 Old_age
Always - 0
225 Load_Cycle_Count 0x0032 097 097 000 Old_age
Always - 40377
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
Note: selective self-test log revision number (0) not 1 implies that no
selective self-test has ever been run
SMART Selective self-test log data structure revision number 0
Note: revision number not 1 implies that no selective self-test has ever
been run
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Completed [00% left] (0-65535)
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad8 - WDC WD20EARS-00S8B1/80.00A80
=== START OF INFORMATION SECTION ===
Model Family: Western Digital Caviar Green (Adv. Format)
Device Model: WDC WD20EARS-00S8B1
Serial Number: ******************
LU WWN Device Id: 5 0014ee 2595827c0
Firmware Version: 80.00A80
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Fri Oct 28 18:11:23 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x84) Offline data collection activity
was suspended by an interrupting command from host.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (43200) seconds.
Offline data collection
capabilities: (0x7b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
Conveyance self-test routine
recommended polling time: ( 5) minutes.
SCT capabilities: (0x3031) SCT Status supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail
Always - 0
3 Spin_Up_Time 0x0027 152 141 021 Pre-fail
Always - 9391
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 562
5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 200 200 000 Old_age
Always - 0
9 Power_On_Hours 0x0032 082 082 000 Old_age
Always - 13428
10 Spin_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 205
192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age
Always - 154
193 Load_Cycle_Count 0x0032 197 197 000 Old_age
Always - 9707
194 Temperature_Celsius 0x0022 120 108 000 Old_age
Always - 32
196 Reallocated_Event_Count 0x0032 200 200 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 200 200 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 200 200 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age
Always - 7
200 Multi_Zone_Error_Rate 0x0008 200 200 000 Old_age
Offline - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
Device /dev/ad10 - WDC WD20EARS-00S8B1/80.00A80
=== START OF INFORMATION SECTION ===
Model Family: Western Digital Caviar Green (Adv. Format)
Device Model: WDC WD20EARS-00S8B1
Serial Number: ***************
LU WWN Device Id: 5 0014ee 2aeadc72d
Firmware Version: 80.00A80
User Capacity: 2,000,398,934,016 bytes [2.00 TB]
Sector Size: 512 bytes logical/physical
Device is: In smartctl database [for details use: -P show]
ATA Version is: 8
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Fri Oct 28 18:11:24 2011 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled
=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
General SMART Values:
Offline data collection status: (0x82) Offline data collection activity
was completed without error.
Auto Offline Data Collection: Enabled.
Self-test execution status: ( 0) The previous self-test routine
completed
without error or no self-test has ever
been run.
Total time to complete Offline
data collection: (42360) seconds.
Offline data collection
capabilities: (0x7b) SMART execute Offline immediate.
Auto Offline data collection on/off support.
Suspend Offline collection upon new
command.
Offline surface scan supported.
Self-test supported.
Conveyance Self-test supported.
Selective Self-test supported.
SMART capabilities: (0x0003) Saves SMART data before entering
power-saving mode.
Supports SMART auto save timer.
Error logging capability: (0x01) Error logging supported.
General Purpose Logging supported.
Short self-test routine
recommended polling time: ( 2) minutes.
Extended self-test routine
recommended polling time: ( 255) minutes.
Conveyance self-test routine
recommended polling time: ( 5) minutes.
SCT capabilities: (0x3031) SCT Status supported.
SCT Feature Control supported.
SCT Data Table supported.
SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED
WHEN_FAILED RAW_VALUE
1 Raw_Read_Error_Rate 0x002f 200 200 051 Pre-fail
Always - 0
3 Spin_Up_Time 0x0027 157 144 021 Pre-fail
Always - 9133
4 Start_Stop_Count 0x0032 100 100 000 Old_age
Always - 529
5 Reallocated_Sector_Ct 0x0033 200 200 140 Pre-fail
Always - 0
7 Seek_Error_Rate 0x002e 200 200 000 Old_age
Always - 0
9 Power_On_Hours 0x0032 082 082 000 Old_age
Always - 13513
10 Spin_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
11 Calibration_Retry_Count 0x0032 100 100 000 Old_age
Always - 0
12 Power_Cycle_Count 0x0032 100 100 000 Old_age
Always - 199
192 Power-Off_Retract_Count 0x0032 200 200 000 Old_age
Always - 158
193 Load_Cycle_Count 0x0032 078 078 000 Old_age
Always - 366543
194 Temperature_Celsius 0x0022 123 107 000 Old_age
Always - 29
196 Reallocated_Event_Count 0x0032 200 200 000 Old_age
Always - 0
197 Current_Pending_Sector 0x0032 200 200 000 Old_age
Always - 0
198 Offline_Uncorrectable 0x0030 200 200 000 Old_age
Offline - 0
199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age
Always - 9
200 Multi_Zone_Error_Rate 0x0008 200 200 000 Old_age
Offline - 0
SMART Error Log Version: 1
No Errors Logged
SMART Self-test log structure revision number 1
No self-tests have been logged. [To run self-tests, use: smartctl -t]
SMART Selective self-test log data structure revision number 1
SPAN MIN_LBA MAX_LBA CURRENT_TEST_STATUS
1 0 0 Not_testing
2 0 0 Not_testing
3 0 0 Not_testing
4 0 0 Not_testing
5 0 0 Not_testing
Selective self-test flags (0x0):
After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
==========================================================================================================
That's the complete listing for all four spinning disk drives.
I've blanked the serial numbers and excluded the card reader slots.
>
>> If the head unload rating life rating was 3,000,000, it wouldn't be
>> an issue but, even taking WD's revised estimate of 600,000 into
>> account, this still represents a possible failure within a mere 12
>> months of use so it's vital that the time-out be raised to something
>> more reasonable (like 300 seconds!) if you want to give the drive a
>> fighting chance to survive the 3 (5?) year warranty period.
>
> I'm surprised they're resorting to head unloading at all, given that
> these drives are meant to respond immediately to requests even in
> low-power mode. Mine (bought in 2009) don't unload. (But maybe the
> controller needs to tell the drive not to do that, and mine does that
> automatically?)
latched in the parked state) so that the track servo controller can be
disabled. Why the hell they chose a time out interval of only 8 seconds,
God alone knows. It should have been quite obvious that the 300,000 unload
cycles limit was likely to be reached within less than 6 months of
operation. Just doubling the interval would most likely have extended the
operational hours per 300,000 unload cycles by a factor of three or more.
Better still if they had chosen a much larger time out default figure of
30 seconds or more, leaving the end user to decide their own 'optimum'
time out value. Hopefully, the newer 3TB drives have this feature disabled
by default (along with a more eco-friendly power consumption regime).
>
>> However, head unloading time out issues aside, it's the strange non
>> eco-green power consumption curve that irks me the most. When I
>> installed those drives, I used them to replace a couple of 7200 rpm
>> 1TB Samsung spinpoints that lived alongside a couple of 1TB Samsung
>> 5400 rpm eco green drives, expecting the 53 watt consumption to fall
>> below the 49 watt mark but all that happened was that the power varied
>> between 48 and 53 watts (infrequently as low as 48, most often 53
>> watts regardless of how long the box was effectively 'idle').
>
> Now that I'd expect from a drive that works like mine. The power
> consumption is reduced when the drives are spun down to a reduced RPM,
> at the cost of increased latency and briefly increased power consumption
> while spinning up (much of which is got back when the drive spins down
> again, so you get a dip in power consumption below even the idling
> figure at that point).
preferring the instant availability of data and, most importantly, the
improved reliability of constant spin (minimal temperature variations).
>
>> As it happened, the controller PCB transplant trick worked just fine
>> and I successfully retrieved all the data. The problem was that I
>> under-estimated the PCB's susceptibility to ESD[2] when it came to
>> refitting it back to the original drive it had been borrowed from and
>> I now have two dead Samsung drives with perfect mechanicals for my
>> trouble.
>
> See, this is why I never touch hardware (well, other than the
> coordination deficits). I seem to be entirely made of static
> electricity: I have blown things simply touching machines' cases...
awareness that I developed over thirty years ago. Even now, I'm only
assuming the problem was an ESD accident since it seems rather extreme of
Samsung to have incorporate an "Only One Byte of The Cherry" limit on
controller board transplantation operations (but, who knows? Anything is
possible given microprocessor control and some flash ram).
>
>> [1] What I'd initially assumed was a 4k sector issue with win2k (even
>> after correctly aligning the partition boundary) looks like it may
>> have had more to do with an e-SATA issue within the win2k box since I
>> saw exactly the same slow and stuttered write performance along with
>> mouse movement hangs when I tested the newly released and
>> re-partitioned 512 byte sectored 2TB eco-green Samsung drive in an
>> e-SATA enclosure.
>
> Mouse movement hangs?! If you weren't swapping really heavily, that
> suggests it was saturating or locking up the bus for a long period of
> time. Gah.
these symptoms, I put two and two together and got Five (must have been
large values of two), assuming it was something to do with it being an
effect of the 4k AF used by the WD20EARS drive I was testing. It hadn't
occurred to me that I could have been witnessing the effects of bad SATA
connector flakiness otherwise I would have transplanted it straight into
the desktop machine and bought a third 2TB Samsung eco green drive to take
its place in the server box.
As things are, I won't get the chance to verify whether or not the 4k AF
is an issue under win2k for some time, now that I've splashed out on that
5K3000 Hitachi drive. Between the extra 1TB added to the server box and
the 1TB upgrade on the desktop machine, I should have enough capacity to
keep me going for another 6 to 8 months, by which time (hopefully), the
current price hike on disk drives should be just another bad memory.
--
Regards JB Good
Johny B Good
Oct 28, 2011, 3:36:00 PM10/28/11
to
On Fri, 28 Oct 2011 15:52:14 +0100, Nix <nix-ra...@esperi.org.uk>
wrote:
Just the head unloading. The drives are set to never spin down.
--
Regards JB Good
wrote:
--
Regards JB Good
0 new messages