freebsd-hackers Digest, Vol 364, Issue 5
freebsd-hac...@freebsd.org
freebsd...@freebsd.org
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
or, via email, send a message with subject or body 'help' to
freebsd-hac...@freebsd.org
You can reach the person managing the list at
freebsd-ha...@freebsd.org
When replying, please edit your Subject line so it is more specific
than "Re: Contents of freebsd-hackers digest..."
Today's Topics:
1. Re: ATA 4K sector issues (Dag-Erling Sm?rgrav)
2. HotCloud '10 Submission Deadline Approaching (Lionel Garth Jones)
3. Re: ATA 4K sector issues (Thiago Damas)
4. Re: How to slow down SATA to 1.5 GBit/s ? (Juergen Lock)
5. Re: How to slow down SATA to 1.5 GBit/s ? (Thomas Schmitt)
6. Re: ATA 4K sector issues (Dag-Erling Sm?rgrav)
----------------------------------------------------------------------
Message: 1
Date: Thu, 18 Mar 2010 13:01:09 +0100
From: Dag-Erling Sm?rgrav <d...@des.no>
Subject: Re: ATA 4K sector issues
To: Matthew Dillon <dil...@apollo.backplane.com>
Cc: freebsd...@freebsd.org
Message-ID: <86y6hpj...@ds4.des.no>
Content-Type: text/plain; charset=utf-8
Matthew Dillon <dil...@apollo.backplane.com> writes:
> We experimented a bit with aligning fdisk (dos slices) by changing the
> sector offset to 2 but I came to the conclusion that it was better to
> do the alignment in disklabel / gpt / whatever higher-level
> partitioner floats your boat and not mess with anything the BIOS uses
> to boot the machine
Not sure what you mean by "changing the sector offset to 2", but the
BIOS doesn't care where partitions start or end. It just obeys the
partition table.
> My recommendation is to use a 1MB physical base alignment. That's
> what I adjusted DragonFly's disklabel64 to do. It's definitely best
> to have the partitioner deal with it instead of having to mess around
> manually because the partitioner can calculate the actual physical
> alignment by querying the kernel's disk subsystem regardless of the
> topology.
The disk system doesn't necessarily know. Disks have been known to lie
about their physical sector size. The best strategy is to consistently
use a large alignment "just in case". Using a 4k or 8k or even 1M
alignment has no negative impact on 512b disks (other than that it
wastes a very small amount of disk space), but for a 4k disk or an SSD,
it can make the difference between "dog slow" and "lightning fast".
> * A variety of media already uses much larger physical block sizes.
> MLC flash uses 128K and SLC uses 64K blocks. See the note below
> on why this matters even though SSDs do write combining.
Some SSDs do RAID0 internally and therefore have larger block sizes.
ISTR the OCZ Vertex is made up of four SATA SSDs with a microcontroller
in front.
DES
--
Dag-Erling Smørgrav - d...@des.no
------------------------------
Message: 2
Date: Thu, 18 Mar 2010 09:11:13 -0700
From: Lionel Garth Jones <l...@usenix.org>
Subject: HotCloud '10 Submission Deadline Approaching
To: freebsd...@freebsd.org
Message-ID: <9D753976-BF33-483F...@usenix.org>
Content-Type: text/plain; charset=US-ASCII; format=flowed
We're writing to remind you that submission deadline for the 2nd USENIX
Workshop on Hot Topics in Cloud Computing (HotCloud '10) program
committee is approaching.
Please submit your work by Tuesday, March 23, 2010.
http://www.usenix.org/hotcloud10/cfpb
Cloud computing has attracted a great deal of attention both from the
research community and from industry. The cloud computing paradigm has
evolved over the years from a basic IT infrastructure (data centers) to
platform as a service (PaaS), and then from software as a service (SaaS)
to complete service enablement on a hosted infrastructure (IaaS). At the
same time, virtualization has emerged as a key enabler for the cloud
computing paradigm. Several challenges arise in the design,
implementation, and deployment of virtualized clouds.
HotCloud will provide a forum for academics as well as practitioners in
the field to share their experience, leverage each other's perspectives,
and identifynew/emerging "hot" trends in this area.
Topics of interest include but are not limited to the following:
* Platform as a service
* Software as a service
* Infrastructure as a service
* Elasticity and availability in a cloud
* Multi-tenancy
* Storage cloud
* Charging models and economics
* Power-efficient ("green") computing for clouds
* Virtual appliance management and composition
* Monitoring, troubleshooting, and failure recovery
* Cloud management and configuration
* Programming models
* Security and privacy in clouds
* New applications for clouds
* Mobile clouds
* Cloud usage scenarios
For more details on the submission process, please see the complete
Call for Papers at:
http://www.usenix.org/hotcloud10/cfpb
We look forward to receiving your submissions!
Erich Nahum, IBM T.J. Watson Research Center
Dongyan Xu, Purdue University
HotCloud '10 Program Co-Chairs
hotcloud...@usenix.org
---------------------------------
Call for Papers
2nd USENIX Workshop on Hot Topics in Cloud Computing (HotCloud '10)
June 22, 2010
Boston, MA, USA
http://www.usenix.org/hotcloud10/cfpb
Submissions Deadline: March 23, 2010
---------------------------------
------------------------------
Message: 3
Date: Thu, 18 Mar 2010 15:46:31 -0300
From: Thiago Damas <tda...@gmail.com>
Subject: Re: ATA 4K sector issues
To: Matthew Dillon <dil...@apollo.backplane.com>
Cc: freebsd...@freebsd.org
Message-ID:
<f8e3d83f1003181146o33...@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
I tested now with 1Mb (2048 blocks) at the begining of disk, and same
behaviour: slow write speed and high disk latency.
On Wed, Mar 17, 2010 at 6:17 PM, Thiago Damas <tda...@gmail.com> wrote:
> I'll try tomorrow more zfs tests, with 1M alignment on begining of disk.
> But I also remember that zfs block size its 128k, but metadata can be of
> dynamic size. And we can use compressed files too.
> There is a sysctl, md_compress, that I turned out in my tests, but not
> working as expected.
> Why using gnop -S 4096 works well?
>
> Thiago
>
>
------------------------------
Message: 4
Date: Thu, 18 Mar 2010 21:17:34 +0100 (CET)
From: Juergen Lock <n...@jelal.kn-bremen.de>
Subject: Re: How to slow down SATA to 1.5 GBit/s ?
To: scdb...@gmx.net
Cc: freebsd...@freebsd.org
Message-ID: <201003182017....@triton8.kn-bremen.de>
In article <1057305...@192.168.2.69> you write:
>Hi,
Hi!
>
>> > I found a similar PR
>> > http://www.mail-archive.com/freebsd...@freebsd.org/msg70510.html
>>
>> Hm thats my post, wrong link? :)
>
>Indeed. I copied the wrong URL from my mail to
>mav. The PR is at
> http://www.freebsd.org/cgi/query-pr.cgi?pr=amd64/144151
>
>This here would match my theory:
>>> cdrecord: Input/output error. test unit ready: scsi sendcmd: cmd timeout after 40.010 (40) s
>>> CDB: 00 00 00 00 00 00
>because the command is sent especially to see
>whether any eventual error condition is gone.
>It should get a reply within milliseconds.
>
Oh hm, I'll leave that question to mav... :)
>
>> > and bothered mav for instructions how to upgrade
>> That's also documented in the handbook, starting with `24.5.2 Staying
>> Stable with FreeBSD'
>
>I read that before asking. It seems a bit
>outdated ("7-STABLE") and it is full of warnings.
>Actually i hardly feel ripe for makeworld.html
>
Oh well... I'd say it's easier than it sounds. I guess if you
feel uncomfortable you could also `practice' using a VM, taking a
snapshot before you start the upgrade...
>I will combine that with my endeavor to install
>another FreeBSD from scratch.
>
..or if you do that you could also use a stable snapshot iso from here:
http://pub.allbsd.org/FreeBSD-snapshots/
(you want RELENG_8 which is the cvs branch for whats called stable/8
in svn.)
>
>> Or maybe burning apps should just invoke hal-disable-polling(1), I
>> suspect that's intended for these kind of things...
>
>Interesting. I'll try to find out whether it
>works. (Often hal stuff on Linux does not work.
>The usual remedy is killall hald-addon-storage.)
>
>First i'll have to learn how to get a X desktop.
>Then i have to see whether hald does any harm.
>
>
>Have a nice day :)
>
>Thomas
Good luck! :)
Juergen
------------------------------
Message: 5
Date: Thu, 18 Mar 2010 23:10:36 +0100
From: "Thomas Schmitt" <scdb...@gmx.net>
Subject: Re: How to slow down SATA to 1.5 GBit/s ?
To: freebsd...@freebsd.org
Message-ID: <1057460...@192.168.2.69>
Hi,
> I guess if you feel uncomfortable you could also
> `practice' using a VM, taking a snapshot before you
> start the upgrade...
I do have backups. :))
(Dumb compressed partition images on DVD when a
rescue Linux was booted. Plus MBR copies.)
> you want RELENG_8
So i am downloading
FreeBSD-8.0-RELENG_8-20100318-JPSNAP-amd64-dvd1.iso
I still have 2 partitions unused on the disk.
For now i can afford to install a real system.
> Good luck! :)
I will cry for help when being stuck.
Have a nice day :)
Thomas
------------------------------
Message: 6
Date: Fri, 19 Mar 2010 12:44:37 +0100
From: Dag-Erling Sm?rgrav <d...@des.no>
Subject: Re: ATA 4K sector issues
To: Thiago Damas <tda...@gmail.com>
Cc: freebsd...@freebsd.org
Message-ID: <86hbocz...@ds4.des.no>
Content-Type: text/plain; charset=utf-8
Thiago Damas <tda...@gmail.com> writes:
> I tested now with 1Mb (2048 blocks) at the begining of disk, and same
> behaviour: slow write speed and high disk latency.
It just occurred to me - if I'm not mistaken, when the disk block size
is larger than either DFLTPHYS (64 kB), the kernel will do a
read-modify-write no matter how much data you write. You should make
sure the filesystem never sees a block size larger than that (e.g. with
a four-disk stripe or a five-disk raid5, the individual disks' block
size should not exceed DFLTPHYS / 4, because the array's block size will
be 4 x the LCM of the individual disks' block sizes)
DES
--
Dag-Erling Smørgrav - d...@des.no
------------------------------
End of freebsd-hackers Digest, Vol 364, Issue 5
***********************************************