opencsw: all I can find is "unstable" software list; standard one?

[David Combs]

Hi there,

I just bought two 2500's (red) ($550 each CHEAP), and
have (well, a friend has) installed latest (from Oracle)
sparc solaris 10 os.

(One question: does that include the old "extra software"
Sun had a disk for? Oracle sent only two disks, in PAPER
holders! (also CHEAP, those people!)?)

Anyway, now comes time to install no not Blastwave (dead?
asleep? coma?) but openCSW.

I go there, opencsw.org, but as far as software listings,
although they mention they have three lists (ancient, current
OK, and "unstable"), the only one I can find is the
"unstable" one.

Is there actually a current, trustable set of software,
or only this "unstable" one (which is probably actually
ok, I guess)?

Here's the interaction with the site:
(top line consists of headings for the columns)


| pkgutilRelease branchesMirrorsSoftware listPackage Statistics
|
| This is the list of Solaris software packages in our 'unstable' branch. To request a Solaris package for something that is not currently here, please use the Package Request Form.
|
|
| Software Version Description
| 389_admin 1.1.30,REV=2013.01.07 The 389 LDAP server Admin Tools
| 389_adminutil_dev 1.1.15,REV=2013.01.07 The enterprise-class Open Source LDAP server, libadminutil, libadmsslutil, development files
| 389_admin_console 1.1.8,REV=2012.01.24 389 Directory Server - Management Console Plugin AdminServer
| ...
| ...
| ...



And here's the list of urls I've (casually, quickly!) viewed thus
far:

http://www.opencsw.org/community/questions/1/how-to-get-started-with-opencsw
http://www.opencsw.org/community/questions/8/unstable-testing-stable-catalogs-which-should-i-use
http://www.opencsw.org/get-it/releases/
http://www.opencsw.org/get-it/packages/ "This is the list of Solaris software packages in our 'unstable' branch. To request a Solaris package for something that is not currently here, please use the Package Request Form."
http://www.opencsw.org/manual/index.html
http://www.opencsw.org/manual/genindex.html
http://www.opencsw.org/manual/for-administrators/index.html
http://www.opencsw.org/manual/for-administrators/introduction.html
http://www.opencsw.org/manual/for-administrators/getting-starated.html
http://www.opencsw.org/manual/for-administrators/configuring-services.html
http://www.opencsw.org/manual/for-administrators/.html
http://www.opencsw.org/manual/for-administrators/.html
http://www.opencsw.org/manual/for-administrators/.html

http://www.opencsw.org/get-it/
http://www.opencsw.org/category/announcements/ VIDEO!



QUESTION: Any ideas on where to find the middle, standard, current
OK-to-use list of software?


QUESTION: Install not totally complete yet (zfs not set
up for mirroring, etc), so I simply ask you this:
They say to use WGET. I assume that that is in fact
part of what's in the solaris install? (for how else
am I to get pkgutil?)

(Of course it comes with cygwin, which is what I'm
using now via a PC)

QUESTION 3: Need I watch opencsw's 1-hour video they
say they have?

(I will say that their written instructions leave much
to be desired; or, perhaps, it's actually wonderful,
if you just knew where to find it!)

QUESTION 4: Any hints on using opencsw? (Never
used it before; prior stuff was from blastwave,
downloaded installed by someone else, not me (as
those who know me might well expect!))


Thanks much!


David

[YTC#1]

On 22/04/13 21:42, David Combs wrote:
> Hi there,
>
> I just bought two 2500's (red) ($550 each CHEAP), and
> have (well, a friend has) installed latest (from Oracle)
> sparc solaris 10 os.

>
> (One question: does that include the old "extra software"

Which poses the question.... what software are you looking for ? S10
(and S11) include just about all the major stuff you need out of the box.

Like wget ....:-)

> Sun had a disk for? Oracle sent only two disks, in PAPER
> holders! (also CHEAP, those people!)?)

Most people download the ISOs and go from there

>
> Anyway, now comes time to install no not Blastwave (dead?
> asleep? coma?) but openCSW.

Again, install what ?

--
Bruce Porter
"The internet is a huge and diverse community but mainly friendly"
http://blog.maui.co.uk/index.php/ytc/
There *is* an alternative! http://www.openoffice.org/

[David Combs]

In article <kl47o4$mv7$1...@dont-email.me>, YTC#1 <b...@ytc1-spambin.co.uk> wrote:
>On 22/04/13 21:42, David Combs wrote:
>> Hi there,
>>
>> I just bought two 2500's (red) ($550 each CHEAP), and
>> have (well, a friend has) installed latest (from Oracle)
>> sparc solaris 10 os.
>
>>
>> (One question: does that include the old "extra software"
>
>Which poses the question.... what software are you looking for ? S10
>(and S11) include just about all the major stuff you need out of the box.

Really? Earlier versions of s10 sure didn't. Had to install
a LOT of (then blastwave) packages.

>
>Like wget ....:-)
>
>> Sun had a disk for? Oracle sent only two disks, in PAPER
>> holders! (also CHEAP, those people!)?)
>
>Most people download the ISOs and go from there

Stupid question: what are "the ISOs"?

>
>>
>> Anyway, now comes time to install no not Blastwave (dead?
>> asleep? coma?) but openCSW.
>
>Again, install what ?

All the stuff we need that doesn't automatically come from sun/oracle
that we need.
Have no list in front of me right now; if you really want,
I can post a list of 30 or so that I want to have.

Question: why would it be on csw's list if solaris
already had it?

Used to be you had to download GNU egrep, etc, since Sun's
was ancient and no good. Well, there's now gegrep, I recall,
but lots of OTHER stuff still needs to be grabbed externally
(ie blastwave or csw).

My problem was not being able to find the csw-advertised
trustable-versions-of-software list, only the one that
they said might or might not work (I forget the word,
is right here in this email somewhere).

PLEASE, how to find ALL THREE of csw's advertised
software list.

[John D Groenveld]

In article <kl47c5$6rq$1...@panix1.panix.com>,

David Combs <dkc...@panix.com> wrote:
>QUESTION: Any ideas on where to find the middle, standard, current
>OK-to-use list of software?

You might try the OpenCSW users mailing list:
<URL:https://lists.opencsw.org/mailman/listinfo/users>

John
groe...@acm.org

[John D Groenveld]

In article <kl4hos$3tj$1...@reader1.panix.com>,

David Combs <dkc...@panix.com> wrote:
>Really? Earlier versions of s10 sure didn't. Had to install
>a LOT of (then blastwave) packages.

Yes, really.
<URL:http://docs.oracle.com/cd/E26505_01/html/E27063/eezdz.html#scrolltoc>

John
groe...@acm.org

[John D Groenveld]

In article <kl4hos$3tj$1...@reader1.panix.com>,
David Combs <dkc...@panix.com> wrote:

>Stupid question: what are "the ISOs"?

<URL:http://en.wikipedia.org/wiki/ISO_image>

>Question: why would it be on csw's list if solaris
>already had it?

Possibly to support previous Solaris versions and
previous Solaris 10 releases.
Or to include more recent package versions than
those that Oracle supports.

John
groe...@acm.org

[David Combs]

In article <kl4jvm$p84q$1...@tr22n12.aset.psu.edu>,

John D Groenveld <groe...@cse.psu.edu> wrote:
>In article <kl4hos$3tj$1...@reader1.panix.com>,
>David Combs <dkc...@panix.com> wrote:
>>Stupid question: what are "the ISOs"?
>
><URL:http://en.wikipedia.org/wiki/ISO_image>
>
>>Question: why would it be on csw's list if solaris
>>already had it?

John: the above is merely a rhetorical (is that the
correct term) question, an answer posed as a question,
to the prior respondent,
Bruce Porter, <b...@ytc1-spambin.co.uk>:

Recall from years ago how Solaris "media kit"
came with various installation disks, plus
an extra one for "optional software" or whatever
it was called. Was to go into /opt/sfw.

And for those for whom that didn't meet all their
needs, there was the then-wonderful (and maybe
still so?) BLASTWAVE, from Dennis Clarke,
with maybe 1,000 packages (.pkg-files) for
downloading, with SO MANY (maybe almost all)
Sun customers downloading/installing the
stuff (in /opt/csw -- was that the dir name?),
that they had to have mirrors galore just about
everywhere.

Then along his once partner-in-Blastwave, name
escapes me right now, who broke away from
Blastwave and started his own Blastwave-like
service, "opencsw" (open csw?).

Anyway, I go to opencsw.org, where they
say they have three versions of the
packages they make available: old (ie too old),
current and reliable for even the most
critical tasks, and then for the truly
brave, the "unstable" version.

What *I* would like to use is their "current"
trustworthy version, certainly NOT their
"unstable" one.

However, when I go to their site, all I
can find, as list of downloadable packages,
is the one titled "unstable".

Perhaps you, John, have used this "opencsw.org"
site, and downloaded/installed packages from
it. If so, then maybe YOU can find this
middle, "safe" set of packages.



Oh, thanks for that url to the oracle page
that showed what software came with Solaris 10.

(Where does it go, /opt/sfw ?)

Thanks!

David

[David Combs]

In article <kl4ipp$b45m$1...@tr22n12.aset.psu.edu>,

John D Groenveld <groe...@cse.psu.edu> wrote:

Thanks for the suffestion, John.


I infer from your two posts thus far that you
likely have NOT EVER used opencsw, nor probably
BLASTWAVE either.

To the other people reading this thread, have
YOU used either of those two repositories?

(a) Which of the two?

(b) And how recently was that?



(Of course there was always sunfreeware, but that
seemed to be keept up to date not so much, compared
to (in days of old, blastwave, and more recently)
opencsw.)

Can any of you supply any other hints re use
of opencsw?

THANKS!

David

[John D Groenveld]

In article <kl4rfk$kps$1...@reader1.panix.com>,

David Combs <dkc...@panix.com> wrote:
>an extra one for "optional software" or whatever
>it was called. Was to go into /opt/sfw.

Solaris Software Companion CD.
Oracle no longer offers it.

>Perhaps you, John, have used this "opencsw.org"
>site, and downloaded/installed packages from
>it. If so, then maybe YOU can find this
>middle, "safe" set of packages.

I used OpenCSW's unstable packages.

John
groe...@acm.org

[Gary R. Schmidt]

On 23/04/2013 12:35 PM, David Combs wrote:
[SNIP]

> I infer from your two posts thus far that you
> likely have NOT EVER used opencsw, nor probably
> BLASTWAVE either.
>
> To the other people reading this thread, have
> YOU used either of those two repositories?
>
> (a) Which of the two?
>
> (b) And how recently was that?

I use OpenCSW, but started with Blastwave and moved to OpenCSW when they
had a bitch fight, and it looked like OpenCSW would be the more likely
to survive. (There is also a local-ish mirror.)

Lately, updates have become so slow, in the "unstable" branch, that I am
thinking about moving to building items from scratch, again.

I will review Blastwave before I do that, and try to work out whether
what I have from them *needs* to be kept up to date.

Frex, the OpenCSW postfix is way, way behind the times - but I've built
that from scratch because I wanted to do things my way, anyway.

Hmm, thought I should do a quick comparison - but www.blastwave.org
isn't loading for me at the moment!

So, use OpenCSW for stuff that you don't want to take the time to build,
or have to keep up-to-date. Leaven that with SunFreeware, because
sometimes that's better, is my current position.

Cheers,
Gary B-)

--
When men talk to their friends, they insult each other.
They don't really mean it.
When women talk to their friends, they compliment each other.
They don't mean it either.

[YTC#1]

On 23/04/13 04:16, John D Groenveld wrote:
> In article <kl4rfk$kps$1...@reader1.panix.com>,
> David Combs <dkc...@panix.com> wrote:
>> an extra one for "optional software" or whatever
>> it was called. Was to go into /opt/sfw.
>
> Solaris Software Companion CD.
> Oracle no longer offers it.

Indeed, but SFW does

http://www.sunfreeware.com/introduction.html

>
>> Perhaps you, John, have used this "opencsw.org"
>> site, and downloaded/installed packages from
>> it. If so, then maybe YOU can find this
>> middle, "safe" set of packages.
>
> I used OpenCSW's unstable packages.
>
> John
> groe...@acm.org
>

[YTC#1]

On 23/04/13 00:40, David Combs wrote:
> In article <kl47o4$mv7$1...@dont-email.me>, YTC#1 <b...@ytc1-spambin.co.uk> wrote:
>> On 22/04/13 21:42, David Combs wrote:
>>> Hi there,
>>>
>>> I just bought two 2500's (red) ($550 each CHEAP), and
>>> have (well, a friend has) installed latest (from Oracle)
>>> sparc solaris 10 os.
>>
>>>
>>> (One question: does that include the old "extra software"
>>
>> Which poses the question.... what software are you looking for ? S10
>> (and S11) include just about all the major stuff you need out of the box.
>
> Really? Earlier versions of s10 sure didn't. Had to install
> a LOT of (then blastwave) packages.

Yes, and times change. A lot of Solaris engineers (and other staff)
wanted those tools. They became part of the OS.

>>
>> Like wget ....:-)
>>
>>> Sun had a disk for? Oracle sent only two disks, in PAPER
>>> holders! (also CHEAP, those people!)?)
>>
>> Most people download the ISOs and go from there
> Stupid question: what are "the ISOs"?

Really ?

Just download them from
http://www.oracle.com/technetwork/server-storage/solaris11/downloads/index.html
(or chose S10 on the tab to the right).

Then either JumpStart using the ISO or cut the ISO to a DVD and install
from that.

Remember ... "The network is the computer" :-)

>
>>
>>>
>>> Anyway, now comes time to install no not Blastwave (dead?
>>> asleep? coma?) but openCSW.
>>
>> Again, install what ?
> All the stuff we need that doesn't automatically come from sun/oracle
> that we need.
> Have no list in front of me right now; if you really want,
> I can post a list of 30 or so that I want to have.

Well, if you don't share we can't answer :-)

I've not had to apply any extras to S10 for many a year.

>
> Question: why would it be on csw's list if solaris
> already had it?

There may well be some things that are still not
update/supported/included in S10 and CSW will host it or a newer version

>
> Used to be you had to download GNU egrep, etc, since Sun's
> was ancient and no good. Well, there's now gegrep, I recall,
> but lots of OTHER stuff still needs to be grabbed externally
> (ie blastwave or csw).

Or via Sunfreeware.

Fine *if* you need it then go get it, don't just get stuff because you
think you need it. Even KSH is now up to ksh93 :-)

>
> My problem was not being able to find the csw-advertised
> trustable-versions-of-software list, only the one that
> they said might or might not work (I forget the word,
> is right here in this email somewhere).


>
> PLEASE, how to find ALL THREE of csw's advertised
> software list.
>>

Have you tried here ?

https://lists.opencsw.org/mailman/listinfo/

[YTC#1]

On 23/04/13 03:35, David Combs wrote:
> In article <kl4ipp$b45m$1...@tr22n12.aset.psu.edu>,
> John D Groenveld <groe...@cse.psu.edu> wrote:
>> In article <kl47c5$6rq$1...@panix1.panix.com>,
>> David Combs <dkc...@panix.com> wrote:
>>> QUESTION: Any ideas on where to find the middle, standard, current
>>> OK-to-use list of software?
>>
>> You might try the OpenCSW users mailing list:
>> <URL:https://lists.opencsw.org/mailman/listinfo/users>
>>
>> John
>> groe...@acm.org
>
> Thanks for the suffestion, John.
>
>
> I infer from your two posts thus far that you
> likely have NOT EVER used opencsw, nor probably
> BLASTWAVE either.
>
> To the other people reading this thread, have
> YOU used either of those two repositories?

Yes

>
> (a) Which of the two?

Both, initially, then OpenCSW after the last of the many splits.

>
> (b) And how recently was that?

Over a year ago, I no longer need either.

I use S11 for my main OS. In any S10 zones if I need something, I go to SFW.

>
>
>
> (Of course there was always sunfreeware, but that
> seemed to be keept up to date not so much, compared
> to (in days of old, blastwave, and more recently)
> opencsw.)
>

[Markus Sonnenberg]

On 4/22/2013 10:42 PM, David Combs wrote:
>
> I go there, opencsw.org, but as far as software listings,
> although they mention they have three lists (ancient, current
> OK, and "unstable"), the only one I can find is the
> "unstable" one.
>
> Is there actually a current, trustable set of software,
> or only this "unstable" one (which is probably actually
> ok, I guess)?

Hi David,

this will give some hints about the directories.
http://mirror.opencsw.org/opencsw/

just trust me and go with the latest unstable catalog.
i'm using the unstable catalog for years now and i had never a problem
with it. but if you want to feel safe then go with "dublin" or "kiel".

regards
markus.

ct,
--
Das Abspringen einer Begrenzungsmauer dient nicht dem direkten
Zurücklegen des Arbeitsweges.
http://www.rz-amper.de

[Doug McIntyre]

"Gary R. Schmidt" <grsc...@acm.org> writes:
>Hmm, thought I should do a quick comparison - but www.blastwave.org
>isn't loading for me at the moment!

Thats because he finally took his toys and went home.
Blastwave is no more.

[Doug McIntyre]

dkc...@panix.com (David Combs) writes:
>Really? Earlier versions of s10 sure didn't. Had to install
>a LOT of (then blastwave) packages.

The software companion disc (when made) for Sol10 covered pretty much
everything I needed. But then again, I also compile most things I need
because the repo's never have anything built the way I want. Solaris11
also has more things available in the repo as well.

>Question: why would it be on csw's list if solaris
>already had it?

I think somebody else answered this, but mostly because Sun's version
was probably older, and some people need to have the latest version
installed due to many reasons.

>Used to be you had to download GNU egrep, etc, since Sun's
>was ancient and no good.

/usr/xpg4/bin/...

Solaris also offered complete 100% backwards compatibility. That can
sometimes be very limiting as well when they've been doing it for so long.

>My problem was not being able to find the csw-advertised
>trustable-versions-of-software list, only the one that
>they said might or might not work (I forget the word,
>is right here in this email somewhere).

You have to read the mailing lists and infer.
I wouldn't say it is "trustable". I'd say the different feeds are
more like, do you want to stick with OpenSSL v.99xxx or do you want
to move to OpenSSL v1.0 with its different default hash algorithm?
Because they changed something fundemental, it is different now,
which could break certain things (ie. making SSL certs for VMware) unless
you know that you have to add certain flags to compensate.

Perhaps you should be compiling your own software that you want.

At the worst case of problems compiling, I may pull what somebody else
has done and apply whatever patches. But usually my main issues is
that this package v5.4 was only tested with that code package v1.2,
and the v1.8 package that is the only one available now changed the
API significantly, so they no longer build together, but the first
package author never bothered updating. Yadda yadda..

[Doug McIntyre]

dkc...@panix.com (David Combs) writes:
>And for those for whom that didn't meet all their
>needs, there was the then-wonderful (and maybe
>still so?) BLASTWAVE, from Dennis Clarke,

Sunfreeware was always the first stop for me.
Blastwave is closed down and gone.

>Then along his once partner-in-Blastwave, name
>escapes me right now, who broke away from
>Blastwave and started his own Blastwave-like
>service, "opencsw" (open csw?).

I think you have the history all wrong. Dennis Clarke was
quite a self-promoter, and took what Phil Brown had written and
started up Blastwave with it. Then the community formed around
Blastwave, but Dennis Clarke is a bit psycho and kept jerking
things around and wouldn't accept that anything but the latest
packages only for the latest Solaris would be acceptable, so
the people that wanted to see more supported packages supported
across many more versions of Solaris started OpenCSW to maintain
that since Sunfreeware was getting kind of stale.

But now Dennis Clarke took his toys and went home.

OpenCSW is a bit disjointed, because like all opensource packages
it tends to be the few maintainers that are interested in that package
have to keep feeding it, and if they drop out, there are holes and
there isn't enough people to just take over something where there is a hole.

[Andreas F. Borchert]

On 2013-04-23, Gary R. Schmidt <grsc...@acm.org> wrote:
> I will review Blastwave before I do that, and try to work out whether
> what I have from them *needs* to be kept up to date.

[..]

> Hmm, thought I should do a quick comparison - but www.blastwave.org
> isn't loading for me at the moment!

Blastwave has been closed last year, see Dennis Clarke's posting
to the pca mailing list from 20 September 2012 [1]:

Seriously, Solaris is dead.
[..]
ps: I should go kill blastwave .. it serves no purpose anymore anyways. yeah .. done

Andreas.

[1] https://lists.univie.ac.at/pipermail/pca/2012-September/003044.html

[David Combs]

In article <qtih4a-...@paranoia.mcleod-schmidt.id.au>,

Gary R. Schmidt <grsc...@acm.org> wrote:

[some pieces snipped out by me]

>
>I use OpenCSW, but started with Blastwave and moved to OpenCSW when they
>had a bitch fight, and it looked like OpenCSW would be the more likely
>to survive. (There is also a local-ish mirror.)
>
>Lately, updates have become so slow, in the "unstable" branch, that I am
>thinking about moving to building items from scratch, again.
>
>

>Frex, the OpenCSW postfix is way, way behind the times - but I've built
>that from scratch because I wanted to do things my way, anyway.

These things you've built from scratch -- how much trouble
would it for you to contribute them to csw?

And, when you build things "your own way", what's the
chance that "your own way" will likely be exactly what
everyone else wants too?

Like, what kind of special-purpose methods are you
talking about, generally, for "your own way".

Different compiler switches? Different make files?

Sure would be a lot easier to be able to download
something that has been built ONCE, compile-errors
fixed, etc, than for EACH OF US to have to duplicate
that same process, error fixing, with maybe not
all of us having the same degree of success.

(Like me: I have no confidence that I'd know
or could even figure out what to do if the
standard "configure && make" failed. Would
sure prefer if a real sys-admin or systems
type did it, for the rest of to use.)

How's Phil doing with csw? Perhaps
anyone who builds something him/herself
could ask Phil if he'd like it.

--

David

[David Combs]

Blastwave and/or/versus openCSW:

Plenty of discussion in the recent past
has shown that it is either impossible
or insanely risky to mix packages
from Blastwave AND opencsw.

(Sunfreeware -- that has no conflict with
either of those two, does it?)

Seems from those discussions that an installation
must choose ONE of those two, and stick 100%
to that.

Yes?

Now, earlier in this thread we read that
Blastwave was basically DEAD.

Also that opencsw was alive, but weak,
but again, NOT DEAD.

Therefore, would it make common sense
for pretty much everyone to choose
opencsw 100% OVER blastwave?

That is, because of the impossibility
of mixing, install NOTHING from blastwave
and download only from either opencsw or
sunfreeware (again, mixing those two
is safe, yes?).

And work with Phil to bring opencsw
up to snuff (if, as hinted at earlier
in this thread, it has many holes,
many way-out-of-date builds, etc.)


---

I would think the effort is worth it.

After all, one can buy refurbished
Suns at like 5% to 10% of their
original price. Lots of Sparc computing
power out there, cheap, with no need
to go crawling to Oracle.

(And who knows what's happening there?)



David

[David Combs]

In article <kl5irb$3n9$1...@dont-email.me>, YTC#1 <b...@ytc1-spambin.co.uk> wrote:
>
>I use S11 for my main OS. In any S10 zones if I need something, I go to SFW.
>
>>
>>
>>
>> (Of course there was always sunfreeware, but that
>> seemed to be keept up to date not so much, compared
>> to (in days of old, blastwave, and more recently)
>> opencsw.)
>>
>
>
>
>--
>Bruce Porter
>"The internet is a huge and diverse community but mainly friendly"
>http://blog.maui.co.uk/index.php/ytc/
>There *is* an alternative! http://www.openoffice.org/

Solaris 11?

How did you get that? By buying a new "Oracle" (sparc) computer?

Thanks,

David

[Casper H.S. Dik]

dkc...@panix.com (David Combs) writes:

>Solaris 11?

>How did you get that? By buying a new "Oracle" (sparc) computer?

Or download it?

Casper

[John D Groenveld]

In article <5188e1e5$0$15878$e4fe...@news2.news.xs4all.nl>,
Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>Or download it?

Yes, Solaris 11.1 can be downloaded from OTN:
<URL:http://www.oracle.com/technetwork/server-storage/solaris11/downloads/>

John
groe...@acm.org

[Doug McIntyre]

dkc...@panix.com (David Combs) writes:
>In article <kl5irb$3n9$1...@dont-email.me>, YTC#1 <b...@ytc1-spambin.co.uk> wrote:
>>
>>I use S11 for my main OS. In any S10 zones if I need something, I go to SFW.

...

>Solaris 11?

>How did you get that? By buying a new "Oracle" (sparc) computer?

Solaris 11 is free to use for "the purpose of developing, testing,
prototyping and demonstrating your applications, and not for any other
purpose."

Commercial/Production use requires you to be on maintenance at the
appropriate license cost from Oracle.

Since you are talking about SPARC machines, it doesn't run on all old
hardware, it only supports fairly modern SPARC systems. Ie. most
UltraSPARC support was removed. It is a bit more permissive about x86
systems.

[Gary R. Schmidt]

On 7/05/2013 2:45 PM, David Combs wrote:
> In article <qtih4a-...@paranoia.mcleod-schmidt.id.au>,
> Gary R. Schmidt <grsc...@acm.org> wrote:
>
> [some pieces snipped out by me]
>
>
>>
>> I use OpenCSW, but started with Blastwave and moved to OpenCSW when they
>> had a bitch fight, and it looked like OpenCSW would be the more likely
>> to survive. (There is also a local-ish mirror.)
>>
>> Lately, updates have become so slow, in the "unstable" branch, that I am
>> thinking about moving to building items from scratch, again.
>>
>>
>> Frex, the OpenCSW postfix is way, way behind the times - but I've built
>> that from scratch because I wanted to do things my way, anyway.
>
> These things you've built from scratch -- how much trouble
> would it for you to contribute them to csw?
>
> And, when you build things "your own way", what's the
> chance that "your own way" will likely be exactly what
> everyone else wants too?

Very unlikely that everyone would want a postfix built with *exactly*
this set of options:
CC=cc
CCARGS="-DUSE_TLS -DUSE_SASL_AUTH -DUSE_CYRUS_SASL -DHAS_DB -I...various"
AUXLIBS="-L and -R and -l this and that"

I don't claim to know postfix well enough to know any more than that was
the set of flags I determined were needed to get a secure authenticated
connection to my ISP's mail server so I could post to more than 10 or 20
recipients at one time.

Ignoring the use of "-R" - which obviously ties things to my machine - I
don't know what the other options require things to be like.

So, no, I don't think that I should contribute this to OpenCSW - perhaps
that's why I built it myself in the first place, the OpenCSW (or
Blastwave, it was a long time ago I started this) version *didn't* have
the secure stuff built in, or something. It was about four years ago
that I did it the first time, I just saved the build scripts and have
applied them to subsequent releases, I didn't make notes.

[Casper H.S. Dik]

Doug McIntyre <mer...@dork.geeks.org> writes:

>Since you are talking about SPARC machines, it doesn't run on all old
>hardware, it only supports fairly modern SPARC systems. Ie. most
>UltraSPARC support was removed. It is a bit more permissive about x86
>systems.

But requires a 64 bit x86 CPU (Solaris 11 removed support for a 32 bit
x86 kernel)

Casper

[Casper H.S. Dik]

"Gary R. Schmidt" <grsc...@acm.org> writes:

>So, no, I don't think that I should contribute this to OpenCSW - perhaps
>that's why I built it myself in the first place, the OpenCSW (or
>Blastwave, it was a long time ago I started this) version *didn't* have
>the secure stuff built in, or something. It was about four years ago
>that I did it the first time, I just saved the build scripts and have
>applied them to subsequent releases, I didn't make notes.

And those who write open source should realize that even open source
distributions want to distribute binaries and "feature flags" should
be run-time options, not compile time options.

Casper

[Rich Teer]

On Tue, 7 May 2013, Casper H.S. Dik wrote:

> And those who write open source should realize that even open source
> distributions want to distribute binaries and "feature flags" should
> be run-time options, not compile time options.

As always, Casper is a voice of reason! I agree with this wholeheartedly.

--
Rich Teer, Publisher
Vinylphile Magazine

www.vinylphilemag.com

[Gary R. Schmidt]

On 11/05/2013 11:55 PM, Rich Teer wrote:
> On Tue, 7 May 2013, Casper H.S. Dik wrote:
>
>> And those who write open source should realize that even open source
>> distributions want to distribute binaries and "feature flags" should
>> be run-time options, not compile time options.
>
> As always, Casper is a voice of reason! I agree with this wholeheartedly.
>

Correct me if I am wrong, but does that mean I have to use /dlopen()/ et
al to access any libraries that I *might* need to use?

Or I have to build it with *everything*[1], and then the end user also
has to install *everything*[2], even if all they want is the basic bits?

If I am correct, that increases the effort needed to get stuff out the
door - and we already have the problem with Open Source being developed
by people who think that _portable_ means "compiles on both RedHat *and*
Ubuntu."

Cheers,
Gary B-)

1 - Everything, in this context, being all possible dependent libraries,
that may not be necessary in all cases. Such as SSL only being used for
secure connections (but I don't need that), Curl only needed if optional
access to a web server is required (why do I have to install the Curl
stuff when I don't hit a web server?), et-bloody-cetera.

2 - Or statically link it with everything - can you spell "bloat"? (Due
to previous stupidity at $ORK, this is one of my current cross-eyed bears.)

[Casper H.S. Dik]

"Gary R. Schmidt" <grsc...@acm.org> writes:

>On 11/05/2013 11:55 PM, Rich Teer wrote:
>> On Tue, 7 May 2013, Casper H.S. Dik wrote:
>>
>>> And those who write open source should realize that even open source
>>> distributions want to distribute binaries and "feature flags" should
>>> be run-time options, not compile time options.
>>
>> As always, Casper is a voice of reason! I agree with this wholeheartedly.

>Correct me if I am wrong, but does that mean I have to use /dlopen()/ et
>al to access any libraries that I *might* need to use?

No; it doesn't mean that. Many libraries are always installed because
you can't have a running system without them.

>Or I have to build it with *everything*[1], and then the end user also
>has to install *everything*[2], even if all they want is the basic bits?

For optional binaries, there are two ways to deal with that:
- create a dependency between your package and the
package containing that binary; this is the course
you must take when the use of that library is not
optional.
- create the use of the binary in an optional module;
see, e.g., do what Apache does.

>If I am correct, that increases the effort needed to get stuff out the
>door - and we already have the problem with Open Source being developed
>by people who think that _portable_ means "compiles on both RedHat *and*
>Ubuntu."

But what do the folks in Red Hat and Ubuntu do when they want to add
that package to their repository?

>1 - Everything, in this context, being all possible dependent libraries,
>that may not be necessary in all cases. Such as SSL only being used for
>secure connections (but I don't need that), Curl only needed if optional
>access to a web server is required (why do I have to install the Curl
>stuff when I don't hit a web server?), et-bloody-cetera.

SSL is not a optional part of Solaris. But see above.

>2 - Or statically link it with everything - can you spell "bloat"? (Due
>to previous stupidity at $ORK, this is one of my current cross-eyed bears.)

Can't do that in Solaris unless you create your own archive libraries
and then later curse yourself when you have to create new binaries for
every security bug fixed in any of those archive libraries.


Casper

[David Combs]

In article <kmaorv$ms2i$1...@tr22n12.aset.psu.edu>,

John D Groenveld <groe...@cse.psu.edu> wrote:

The word originally that you couldn't get 11 unless you
purchased a NEW machine FROM ORACLE.

So, if we can all get this ver 11, does that mean that
it will work on OLD machines, like the sunBlade 2500?

Or is some code built into 11 that looks for
some special new, I WAS BUILT BY ORACLE, and
without that, 11 just won't work AT ALL?

Thanks!,

David

[David Combs]

In article <s-mdnasCBv2ycBXM...@giganews.com>,

I just reasked this question, not have yet gotten to this
post in my traversal through the thread.

So, Blade 2500 red -- would 11 work ok on that?

Thanks!

David

[Boris Racle]

On 2013-05-14, David Combs <dkc...@panix.com> wrote:
> In article <kmaorv$ms2i$1...@tr22n12.aset.psu.edu>,
> John D Groenveld <groe...@cse.psu.edu> wrote:
>>In article <5188e1e5$0$15878$e4fe...@news2.news.xs4all.nl>,
>>Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>>>Or download it?
>>
>>Yes, Solaris 11.1 can be downloaded from OTN:
>><URL:http://www.oracle.com/technetwork/server-storage/solaris11/downloads/>
>>
>>John
>>groe...@acm.org
>
> The word originally that you couldn't get 11 unless you
> purchased a NEW machine FROM ORACLE.

Nobody ever said that.

>
> So, if we can all get this ver 11, does that mean that
> it will work on OLD machines, like the sunBlade 2500?

No. It will work on newer SPARC architecture.

>
> Or is some code built into 11 that looks for
> some special new, I WAS BUILT BY ORACLE, and
> without that, 11 just won't work AT ALL?

No, they pulled out all the support for EOL SPARC boxes.
Show them the MONEY!

You can download it all day, every day! But it won't run on anything that
doesn't make them MONEY!

>
> Thanks!,
>
> David
>

[Boris Racle]

On 2013-05-14, David Combs <dkc...@panix.com> wrote:

> In article <s-mdnasCBv2ycBXM...@giganews.com>,
> Doug McIntyre <mer...@dork.geeks.org> wrote:
>>dkc...@panix.com (David Combs) writes:
>>>In article <kl5irb$3n9$1...@dont-email.me>, YTC#1 <b...@ytc1-spambin.co.uk> wrote:
>>>>
>>

Did you read this part?

>>Since you are talking about SPARC machines, it doesn't run on all old
>>hardware, it only supports fairly modern SPARC systems. Ie. most
>>UltraSPARC support was removed. It is a bit more permissive about x86
>>systems.
>
> I just reasked this question, not have yet gotten to this
> post in my traversal through the thread.
>
> So, Blade 2500 red -- would 11 work ok on that?

Nope!

Show them the MONEY!
Yeah baby YEAH!

>
> Thanks!
>
> David
>
>
>

[hume.sp...@bofh.ca]

David Combs <dkc...@panix.com> wrote:
> The word originally that you couldn't get 11 unless you
> purchased a NEW machine FROM ORACLE.

I don't recall that ever being said. But the support list for 11 is
considerably pared down, and you can't run Solaris 11 *in production*,
on new hardware or old hardware, unless you have a support contract.
Testing and development for free is still okay (you get no updates).

A OS support contact from Oracle is, for all intents and purposes, an OS
licensing fee.

> So, if we can all get this ver 11, does that mean that
> it will work on OLD machines, like the sunBlade 2500?

You might get lucky, but signs point to no. Grab the ISO and give it a
boot and see what happens. I suspect the box may boot, but you'll be
lacking drivers, particularly for the framebuffer, if you use one.

> Or is some code built into 11 that looks for
> some special new, I WAS BUILT BY ORACLE, and
> without that, 11 just won't work AT ALL?

No, you're thinking of Apple. Oracle will simply let benign neglect do the
work for them.

To be blunt, Solaris is a dead OS for general computing. You may not be
doing yourself any favours trying too hard to make 11 work on your Blade.
You're digging yourself into technical debt... you might want to investigate
Illumos/OpenIndiana for SPARC for the moment (Martin Bochnig put a heroic -
perhaps even completely irrational - amount of work into making it work),
or just stick with Solaris 10 until such time as you can ditch SPARC
completely.

I'm a huge SPARC fan, I take no joy in making that recommendation.

--
Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/

[Doug McIntyre]

Boris Racle <bor...@snoracle.com> writes:
>On 2013-05-14, David Combs <dkc...@panix.com> wrote:
>> The word originally that you couldn't get 11 unless you
>> purchased a NEW machine FROM ORACLE.

>Nobody ever said that.

I never heard that either.

>> Or is some code built into 11 that looks for
>> some special new, I WAS BUILT BY ORACLE, and
>> without that, 11 just won't work AT ALL?

>No, they pulled out all the support for EOL SPARC boxes.
>Show them the MONEY!

I do think the support was pulled a little early, but as a company,
I can totally see why they wouldn't want to develop and especially
regression test everything on 12-18 year old hardware (ie. the range
of the UltraSPARC processors).
The first Niagara boxes are now even 8 years old.

They are far from the only company that stop developing for ancient
hardware. Ie. I'm frustrated at Cisco for EOL'ing their 7200 routers,
which have reached end of new support.

[Andrew Gabriel]

In article <IIOdnYTKDblJxQ_M...@giganews.com>,

Doug McIntyre <mer...@dork.geeks.org> writes:
> Boris Racle <bor...@snoracle.com> writes:
>>On 2013-05-14, David Combs <dkc...@panix.com> wrote:
>>> The word originally that you couldn't get 11 unless you
>>> purchased a NEW machine FROM ORACLE.
>
>>Nobody ever said that.
>
> I never heard that either.
>
>>> Or is some code built into 11 that looks for
>>> some special new, I WAS BUILT BY ORACLE, and
>>> without that, 11 just won't work AT ALL?
>
>>No, they pulled out all the support for EOL SPARC boxes.
>>Show them the MONEY!
>
> I do think the support was pulled a little early, but as a company,

They can run Solaris 10, which is on Premier support until 2018,
and Extended support until 2021, and Sustaining support thereafter.
I wonder how many other companies offer such a long support period?

If they hadn't been EOF'ed at that point, they would end up being
supported even longer, for whatever the Solaris 11 support dates
will be.

> I can totally see why they wouldn't want to develop and especially
> regression test everything on 12-18 year old hardware (ie. the range
> of the UltraSPARC processors).

In particular, those systems are missing some hardware virtual memory
features which VM2 uses in Solaris 11. To continue supporting them in
Solaris 11 would have required having a different VM2 implementation
for them, which would need to be taken into account for every later
change in the VM subsystem throughout the life of Solaris 11.

> The first Niagara boxes are now even 8 years old.
>
> They are far from the only company that stop developing for ancient
> hardware. Ie. I'm frustrated at Cisco for EOL'ing their 7200 routers,
> which have reached end of new support.

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]

[John D Groenveld]

In article <kmtk9n$rqr$1...@dont-email.me>, <hume.sp...@bofh.ca> wrote:
>David Combs <dkc...@panix.com> wrote:
>> The word originally that you couldn't get 11 unless you
>> purchased a NEW machine FROM ORACLE.
>
>I don't recall that ever being said. But the support list for 11 is

He's lying.

>> So, if we can all get this ver 11, does that mean that
>> it will work on OLD machines, like the sunBlade 2500?
>
>You might get lucky, but signs point to no. Grab the ISO and give it a

The HCL is here:
<URL:http://www.oracle.com/webfolder/technetwork/hcl/data/s11ga/systems/views/oracle_systems_all_results.mfg.page1.html>

>To be blunt, Solaris is a dead OS for general computing. You may not be
>doing yourself any favours trying too hard to make 11 work on your Blade.
>You're digging yourself into technical debt... you might want to investigate
>Illumos/OpenIndiana for SPARC for the moment (Martin Bochnig put a heroic -
>perhaps even completely irrational - amount of work into making it work),
>or just stick with Solaris 10 until such time as you can ditch SPARC
>completely.

Kudos Martin!
<URL:http://wiki.openindiana.org/oi/MartUX_OpenIndiana+oi_151a+SPARC+LiveDVD>

John
groe...@acm.org

[David Combs]

In article <qtih4a-...@paranoia.mcleod-schmidt.id.au>,
Gary R. Schmidt <grsc...@acm.org> wrote:

>On 23/04/2013 12:35 PM, David Combs wrote:
>[SNIP]
>> I infer from your two posts thus far that you
>> likely have NOT EVER used opencsw, nor probably
>> BLASTWAVE either.
>>
>> To the other people reading this thread, have
>> YOU used either of those two repositories?
>>
>> (a) Which of the two?
>>
>> (b) And how recently was that?

>
>I use OpenCSW, but started with Blastwave and moved to OpenCSW when they
>had a bitch fight, and it looked like OpenCSW would be the more likely
>to survive. (There is also a local-ish mirror.)
>
>Lately, updates have become so slow, in the "unstable" branch, that I am
>thinking about moving to building items from scratch, again.
>

>I will review Blastwave before I do that, and try to work out whether
>what I have from them *needs* to be kept up to date.
>

>Frex, the OpenCSW postfix is way, way behind the times - but I've built
>that from scratch because I wanted to do things my way, anyway.

For those things that you yourself build (or have built somewhat
recently) where "your way" is likely to be *everyone's" way,
PLEASE, PLEASE update openCSW (talk to Phil <last Name>, who runs
that site, who is the one who was in the fight between opencsw
and Dennis Clarke (who IS (was?) Mr. Blastwave, at blas...@gmail.com)).

THOUSANDS of solaris-sparc users will thank you, either directly
or via ESP, if you do!

>
>Hmm, thought I should do a quick comparison - but www.blastwave.org
>isn't loading for me at the moment!
>

>So, use OpenCSW for stuff that you don't want to take the time to build,
>or have to keep up-to-date. Leaven that with SunFreeware, because
>sometimes that's better, is my current position.
>
> Cheers,
> Gary B-)
>

[David Combs]

In article <IIOdnYTKDblJxQ_M...@giganews.com>,

QUESTION: if the newest version of 10 (from Oracle, this latest
version of 10, I think) works on NON-oracle hardware (and
why wouldn't it, unless they deliberately made it NOT work
on non-Oracle-manufactured hardware), then shouldn't 11
also work on those same machines? (Again, unless deliberately
made to NOT work there.)


--- BUSINESS ADVICE TO ORACLE: (self serving, but still valid):


Do whatever is necessary to MAKE new versions work on old
Sun machines. WHY?

To eliminate a (the main?) reason for users of old suns
to HATE that guy who runs Oracle.

Any high-up sys-admin who HATES that guy is very unlikely
to push his company to buy ANY hardware from Oracle; rather,
they'll recommend x86, not just because of its low
price but that enhanced by the HATE towards anything Oracle.

Further, that hate gets propagated around at conferences,
sysadmin listserves, and places like this newsgroup.

QUESTION: How many of you solaris users are, just because
you cannot get 11 (and 12, ...) to work on your
current non-Oracle hardware, going to now go BUY
some ORACLE hardware? 10? 5? None of you?

David

[David Combs]

In article <kmu1q8$e5j$1...@dont-email.me>,

Andrew Gabriel <and...@cucumber.demon.co.uk> wrote:
>In article <IIOdnYTKDblJxQ_M...@giganews.com>,
> Doug McIntyre <mer...@dork.geeks.org> writes:
>> Boris Racle <bor...@snoracle.com> writes:
>>>On 2013-05-14, David Combs <dkc...@panix.com> wrote:
>>>> The word originally that you couldn't get 11 unless you
>>>> purchased a NEW machine FROM ORACLE.
>>
>>>Nobody ever said that.
>>
>> I never heard that either.
>>
>>>> Or is some code built into 11 that looks for
>>>> some special new, I WAS BUILT BY ORACLE, and
>>>> without that, 11 just won't work AT ALL?
>>
>>>No, they pulled out all the support for EOL SPARC boxes.
>>>Show them the MONEY!
>>
>> I do think the support was pulled a little early, but as a company,
>
>They can run Solaris 10, which is on Premier support until 2018,
>and Extended support until 2021, and Sustaining support thereafter.
>I wonder how many other companies offer such a long support period?

Are any of these Solaris 10 upgrades available FREE?

(Am a one-man shop; with down economy, there's NO WAY
I can affort a support-contract with Oracle!)


Thanks,

David

[Ian Collins]

David Combs wrote:
> In article <kmu1q8$e5j$1...@dont-email.me>,
> Andrew Gabriel <and...@cucumber.demon.co.uk> wrote:
>> In article <IIOdnYTKDblJxQ_M...@giganews.com>,
>> Doug McIntyre <mer...@dork.geeks.org> writes:
>>> Boris Racle <bor...@snoracle.com> writes:
>>>> On 2013-05-14, David Combs <dkc...@panix.com> wrote:
>>>>> The word originally that you couldn't get 11 unless you
>>>>> purchased a NEW machine FROM ORACLE.
>>>
>>>> Nobody ever said that.
>>>
>>> I never heard that either.
>>>
>>>>> Or is some code built into 11 that looks for
>>>>> some special new, I WAS BUILT BY ORACLE, and
>>>>> without that, 11 just won't work AT ALL?
>>>
>>>> No, they pulled out all the support for EOL SPARC boxes.
>>>> Show them the MONEY!
>>>
>>> I do think the support was pulled a little early, but as a company,
>>
>> They can run Solaris 10, which is on Premier support until 2018,
>> and Extended support until 2021, and Sustaining support thereafter.
>> I wonder how many other companies offer such a long support period?
>
>
> Are any of these Solaris 10 upgrades available FREE?

Do you ever bother to check things for yourself before posting?

--
Ian Collins

[hume.sp...@bofh.ca]

David Combs <dkc...@panix.com> wrote:
> Any high-up sys-admin who HATES that guy is very unlikely
> to push his company to buy ANY hardware from Oracle; rather,
> they'll recommend x86, not just because of its low
> price but that enhanced by the HATE towards anything Oracle.

I don't think you're aware at all how purchasing decisions are made in the
companies Oracle cares about.

With all the *other* stuff Oracle management does on a regular basis, you
really think dropping support for decade-old workstations is going to be the
action that causes an exodus?

[Casper H.S. Dik]

dkc...@panix.com (David Combs) writes:

>QUESTION: if the newest version of 10 (from Oracle, this latest
>version of 10, I think) works on NON-oracle hardware (and
>why wouldn't it, unless they deliberately made it NOT work
>on non-Oracle-manufactured hardware), then shouldn't 11
>also work on those same machines? (Again, unless deliberately
>made to NOT work there.)

Solaris 10 and Solaris 11 support different platforms; in
every Solaris release, support for new hardware is added
and support for older hardware is removed.

In Solaris 11, support for UltraSPARC with sun4u kernel
architecture was removed and support for 32 bit Intel
CPUs.

Support for really old hardware makes little or no sense;
typically, we've noticed that systems are installed with
a recent OS version and that version is generally never
updated.

To keep software working on older hardware costs money:
code that isn't tested will rot, so testing is required
and fixing the software to work on older systems also
cost money (some of the tests will fail).

Casper

[Stefan Krueger]

Talking about rotting code...

Does anyone at oracle tests cpustat these days? As I'm able to easily
panic a Solaris 11.1 (x86 + really new hardware Ivy Bridge CPU ;))
system by running:

# cpustat -c PAPI_tlb_dm 3 5

(this even works in a VM)

Panic string is:

root@solarized:~# fmdump -Vp -u 58d6eb09-14d8-452f-c515-aee4f3ac4a82
TIME UUID
SUNW-MSG-ID
Apr 03 2013 21:18:13.401595000 58d6eb09-14d8-452f-c515-aee4f3ac4a82
SUNOS-8000-KL

TIME CLASS ENA
Apr 03 21:18:00.7657 ireport.os.sunos.panic.dump_available
0x0000000000000000
Apr 03 21:17:59.5319 ireport.os.sunos.panic.dump_pending_on_device
0x0000000000000000

nvlist version: 0
version = 0x0
class = list.suspect
uuid = 58d6eb09-14d8-452f-c515-aee4f3ac4a82
code = SUNOS-8000-KL
diag-time = 1365016693 396205
de = fmd:///module/software-diagnosis
fault-list-sz = 0x1
__case_state = 0x1
topo-uuid = a50a7672-beb8-6d92-8e1e-82401b0958f7
fault-list = (array of embedded nvlists)
(start fault-list[0])
nvlist version: 0
version = 0x0
class = defect.sunos.kernel.panic
certainty = 0x64
asru =
sw:///:path=/var/crash/.58d6eb09-14d8-452f-c515-aee4f3ac4a82
resource =
sw:///:path=/var/crash/.58d6eb09-14d8-452f-c515-aee4f3ac4a82
savecore-succcess = 1
dump-dir = /var/crash
dump-files = vmdump.0
os-instance-uuid =
58d6eb09-14d8-452f-c515-aee4f3ac4a82
panicstr = BAD TRAP: type=e (#pf Page fault)
rp=fffffffc81a50940 addr=0 occurred in module "pcbe.GenuineIntel.6"
due to a NULL pointer dereference
panicstack = unix:die+105 () | unix:trap+153e () |
unix:cmntrap+e6 () | pcbe.GenuineIntel.6:ptm_pcbe_configure+394 () |
genunix:kcpc_configure_reqs+c1 () | genunix:kcpc_bind_cpu+79 () |
cpc:kcpc_ioctl+1b7 () | genunix:cdev_ioctl+6e () |
specfs:spec_ioctl+5d () | genunix:fop_ioctl+d6 () | genunix:ioctl+188
() | unix:brand_sys_sysenter+1dc () |
crashtime = 1365016637
panic-time = April 3, 2013 09:17:17 PM CEST CEST
(end fault-list[0])

Regards,

Stefan

[Casper H.S. Dik]

Stefan Krueger <stadt...@gmx.de> writes:

>Talking about rotting code...

>Does anyone at oracle tests cpustat these days? As I'm able to easily
>panic a Solaris 11.1 (x86 + really new hardware Ivy Bridge CPU ;))
>system by running:

Yes. (What exactly type of hardware?)

># cpustat -c PAPI_tlb_dm 3 5

>(this even works in a VM)

It panics the VM or it panics the host running the VM?


Did you report a bug at Oracle?

Casper

[John D Groenveld]

In article <519b6535$0$15979$e4fe...@news2.news.xs4all.nl>,

Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:

>Yes. (What exactly type of hardware?)

VirtualBox 4.2.12

>It panics the VM or it panics the host running the VM?

Panics the guest where the guest is running Solaris 11.1 SRU 7.5.

>Did you report a bug at Oracle?

No.

John
groe...@acm.org

[Casper H.S. Dik]

groe...@cse.psu.edu (John D Groenveld) writes:

>In article <519b6535$0$15979$e4fe...@news2.news.xs4all.nl>,
>Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>>Yes. (What exactly type of hardware?)

>VirtualBox 4.2.12

I meant the underlying Intel hardware.

Casper

[John D Groenveld]

In article <519b6d09$0$15863$e4fe...@news2.news.xs4all.nl>,

Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>I meant the underlying Intel hardware.

Intel Core i7-3610QM

John
groe...@acm.org

[Stefan Krueger]

Hi,

Type of hardware? I think all you need is an Intel Ivy Bridge CPU
(tried a i5-3470 and a i7-3615QM so far).

I did not report a bug to Oracle because all the machines I have are
not officially supported (the x86 HCL is also rather thin...).

Also what John said, it panics the VM, not the whole system (but I
must admin that I didn't try it on a Solaris host running solaris
inside VBox...)

Another (rather inconvenient...) panic can be triggered with (as seen
on freenode's #solaris irc chan):

root@solarized:/dev# ls -l null
lrwxrwxrwx 1 root root 27 Apr 11 00:06 null ->
../devices/pseudo/mm@0:null
root@solarized:/dev# ln -s /devices/pseudo/mm\@0\:null null2
root@solarized:/dev# mv null2 null

panicstr = assertion failed: vp->v_count > 0, file:
../../common/fs/vnode.c, line: 1299
panicstack = fffffffffbabd959 () | genunix:vn_rele+79
() | genunix:vn_renameat+3dc () | genunix:renameat+86 () |
unix:brand_sys_sysenter+1dc () |

PS: Were would mere mortals without a support contract report such
things?

[John D Groenveld]

In article <knfr6j$4be$1...@dont-email.me>,

Stefan Krueger <stadt...@gmx.de> wrote:
>PS: Were would mere mortals without a support contract report such
>things?

Here is great.

John
groe...@acm.org

[Casper H.S. Dik]

groe...@cse.psu.edu (John D Groenveld) writes:

>In article <519b6d09$0$15863$e4fe...@news2.news.xs4all.nl>,
>Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>>I meant the underlying Intel hardware.

>Intel Core i7-3610QM

And did it also panic natively?

Casper

[John D Groenveld]

In article <519b78fc$0$15874$e4fe...@news2.news.xs4all.nl>,

Casper H.S. Dik <Caspe...@OrSPaMcle.COM> wrote:
>And did it also panic natively?

No.
That event is not supported when Solaris 11.1 is running
natively.

John
groe...@acm.org

[Stefan Krueger]

On 2013-05-21, John D Groenveld <groe...@cse.psu.edu> wrote:
> In article <knfr6j$4be$1...@dont-email.me>,
> Stefan Krueger <stadt...@gmx.de> wrote:
>>PS: Were would mere mortals without a support contract report such
>>things?
>
> Here is great.

Good, since I'm allowed to bitch here, how about this one:

I intended to run a apache (in worker mode instead of prefork)
webserver with php using the fcigd module, but this horribly fails:

# pkg install apache-22 apache-fcgid php-53 apache-php53
# /usr/php/5.3/bin/php-cgi -i > /dev/null
ld.so.1: php-cgi: fatal: relocation error: file
/usr/php/5.3/modules/pdo_sqlite.so: symbol sqlite3_libversion:
referenced symbol not found
Killed
# /usr/php/5.3/bin/php-cgi -c /etc/php/5.3/nsapi/php.ini -i > /dev/null
PHP Warning: PHP Startup: Unable to load dynamic library
'/usr/php/5.3/zts-modules/bz2.so' - ld.so.1: php-cgi: fatal:
relocation error: file /usr/php/5.3/zts-modules/bz2.so: symbol
core_globals_id: referenced symbol not found in Unknown on line 0
PHP Warning: PHP Startup: Unable to load dynamic library
'/usr/php/5.3/zts-modules/curl.so' - ld.so.1: php-cgi: fatal:
relocation error: file /usr/php/5.3/zts-modules/curl.so: symbol
executor_globals_id: referenced symbol not found in Unknown on line 0
PHP Warning: PHP Startup: ftp: Unable to initialize module
Module compiled with build ID=API20090626,TS
PHP compiled with build ID=API20090626,NTS
These options need to match
in Unknown on line 0
PHP Warning: PHP Startup: Unable to load dynamic library
'/usr/php/5.3/zts-modules/gd.so' - ld.so.1: php-cgi: fatal: relocation
error: file /usr/php/5.3/zts-modules/gd.so: symbol core_globals_id:
referenced symbol not found in Unknown on line 0
PHP Warning: PHP Startup: gettext: Unable to initialize module
Module compiled with build ID=API20090626,TS
PHP compiled with build ID=API20090626,NTS
These options need to match
in Unknown on line 0
etc.

Also:

# ls /usr/php/5.3/*modu*/sqlit*
/usr/php/5.3/modules/sqlite.so
/usr/php/5.3/zts-modules/sqlite3.so
# grep '.' /etc/php/5.3/*conf.d/sqlite*
/etc/php/5.3/conf.d/sqlite.ini:extension=sqlite.so
/etc/php/5.3/zts-conf.d/sqlite.ini:extension=sqlite.so

(Shouldn't that be sqlite3.so in zts's sqlite.ini?)

Regards,

Stefan

[Casper H.S. Dik]

Stefan Krueger <stadt...@gmx.de> writes:

>Talking about rotting code...

>Does anyone at oracle tests cpustat these days? As I'm able to easily
>panic a Solaris 11.1 (x86 + really new hardware Ivy Bridge CPU ;))
>system by running:

># cpustat -c PAPI_tlb_dm 3 5

>(this even works in a VM)

It seems to "only work in a VM" not "just". I have several
systems which allow me to run this command and only on
Virtual Box it panics. I did not find the bug so I filed it.

Casper

[Casper H.S. Dik]

Stefan Krueger <stadt...@gmx.de> writes:

>Another (rather inconvenient...) panic can be triggered with (as seen
>on freenode's #solaris irc chan):

>root@solarized:/dev# ls -l null
>lrwxrwxrwx 1 root root 27 Apr 11 00:06 null ->
>../devices/pseudo/mm@0:null
>root@solarized:/dev# ln -s /devices/pseudo/mm\@0\:null null2
>root@solarized:/dev# mv null2 null

This is a known issue.

Casper

[Casper H.S. Dik]

groe...@cse.psu.edu (John D Groenveld) writes:

It depends on the hardware; on an Atom box I own
defines that event but doesn't panic when running
cpustat.

I'm surprised that Virtual Box adds events over the
ones defined by the underlying hardware.

Casper