From stefan at mdy.univie.ac.at  Tue Sep  1 08:01:29 2009
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Tue, 1 Sep 2009 17:01:29 +0200
Subject: [Caos] jmicron module missing during install
Message-ID: <20090901150129.GB10349@loop.mdy.univie.ac.at>

The following pertains to an install using

caos-nsa-full-1.0.8.x86_64.iso

(md5sum checked OK. I was not able to find a later install CD, let alone
the live CD described in the wiki ???)

The problem arose on rather new machines, Intel Xeon Nehalem (E5530/X5550)
and matching supermicro motherboards.  Since there were only ATA CD/DVD
drives on these machines, the jmicron stuff is used to simulate SCSI.

Unfortunately, Caos NSA 1.0.8 installer doesn't install the jmicron
module, and so while the "test" target reports SUCCESS, the
"autoinstall" target fails. To add insult to injury, the jmicron
module is included on the boot CD.  A naive use of the (by now
undocumented(?)) preshell boot option doesn't help, since the
check for the presence of the cinch installer is carried out before
the preshell is executed, so loading jmicron in the preshell is too
late ...

I finally could install after studying the init scripts on the CD and
manually carrying out *again* the check for the cinch installer after
having loaded the jmicron module. So, while this problem is solved, I thought
this might help someone later on. At present, this raises the following
issues.

* Is it only stupid me, or describes the wiki something I couldn't find to
download. Presumably, with a live cd, the above issue could be solved easily.

* Why is the preshell option not documented anymore. I knew about it
because I needed it in the past to manually load an ethernet driver;
it was docemnted then. This time I had to read the init script to
remember the exact "incantation"

* The specific problem suggests for me that preshell (if specified)
should be the first thing to be executed. It's the logical point to
fix things that the autodetection missed. Checking for the cinch
installer should come later ...

* Having "test" report success only to be greated with 'cannot find
cinch installer, rebooting in 60 seconds' when attempting
"autoinstall" resulted in some medium to heavy cursing ... (one might
consider finding the cinch installer a prerequisite for success ?)

[Don't construe this as too much of a rant; caos never claimed to be for
aunt Tilly ... in retrospect reading through these nash scripts was quite
educational ;-) ]

Cheers,

Stefan

-- 
Stefan Boresch
Institute for Computational Biological Chemistry
University of Vienna, Waehringerstr. 17       A-1090 Vienna, Austria
Phone: -43-1-427752715                        Fax:   -43-1-427752790

From gmkurtzer at gmail.com  Tue Sep  1 09:05:25 2009
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Tue, 1 Sep 2009 09:05:25 -0700
Subject: [Caos] jmicron module missing during install
In-Reply-To: <20090901150129.GB10349@loop.mdy.univie.ac.at>
References: <20090901150129.GB10349@loop.mdy.univie.ac.at>
Message-ID: <571f1a060909010905t6435e8dexf785e9e81d002b17@mail.gmail.com>

Hi Stefan.

Comments inline...

On Tue, Sep 1, 2009 at 8:01 AM, Stefan Boresch<stefan at mdy.univie.ac.at> wrote:
> The following pertains to an install using
>
> caos-nsa-full-1.0.8.x86_64.iso
>
> (md5sum checked OK. I was not able to find a later install CD, let alone
> the live CD described in the wiki ???)
>
> The problem arose on rather new machines, Intel Xeon Nehalem (E5530/X5550)
> and matching supermicro motherboards. ?Since there were only ATA CD/DVD
> drives on these machines, the jmicron stuff is used to simulate SCSI.
>
> Unfortunately, Caos NSA 1.0.8 installer doesn't install the jmicron
> module, and so while the "test" target reports SUCCESS, the
> "autoinstall" target fails. To add insult to injury, the jmicron
> module is included on the boot CD. ?A naive use of the (by now
> undocumented(?)) preshell boot option doesn't help, since the
> check for the presence of the cinch installer is carried out before
> the preshell is executed, so loading jmicron in the preshell is too
> late ...
>
> I finally could install after studying the init scripts on the CD and
> manually carrying out *again* the check for the cinch installer after
> having loaded the jmicron module. So, while this problem is solved, I thought
> this might help someone later on. At present, this raises the following
> issues.
>
> * Is it only stupid me, or describes the wiki something I couldn't find to
> download. Presumably, with a live cd, the above issue could be solved easily.

Yes, we know. We were planning on releasing the live media installer
already, but then there were several high priority exploits that were
found in the current kernels.

For testing purposes, they can be found at:
http://altruistic.infiscale.org/isos. Please expect a new one in the
next day or so, then we will push that to the live mirrors.

>
> * Why is the preshell option not documented anymore. I knew about it
> because I needed it in the past to manually load an ethernet driver;
> it was docemnted then. This time I had to read the init script to
> remember the exact "incantation"

It was documented at one point in the wiki, but I am not sure what
happened to it or if someone removed it.

>
> * The specific problem suggests for me that preshell (if specified)
> should be the first thing to be executed. It's the logical point to
> fix things that the autodetection missed. Checking for the cinch
> installer should come later ...

Agreed.

>
> * Having "test" report success only to be greated with 'cannot find
> cinch installer, rebooting in 60 seconds' when attempting
> "autoinstall" resulted in some medium to heavy cursing ... (one might
> consider finding the cinch installer a prerequisite for success ?)

Yes, again I agree.

>
> [Don't construe this as too much of a rant; caos never claimed to be for
> aunt Tilly ... in retrospect reading through these nash scripts was quite
> educational ;-) ]

Since we are moving the installer to the live media solution I don't
think most of these issues will continue to be a problem. If you have
the opportunity, I would be very interested in some feedback on the
new live media installer.

Thanks,
Greg

-- 
Greg Kurtzer
http://www.infiscale.com/
http://www.perceus.org/
http://www.caoslinux.org/

From stefan at mdy.univie.ac.at  Wed Sep  2 00:19:22 2009
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 2 Sep 2009 09:19:22 +0200
Subject: [Caos] jmicron module missing during install
In-Reply-To: <571f1a060909010905t6435e8dexf785e9e81d002b17@mail.gmail.com>
References: <20090901150129.GB10349@loop.mdy.univie.ac.at>
	<571f1a060909010905t6435e8dexf785e9e81d002b17@mail.gmail.com>
Message-ID: <20090902071922.GC10349@loop.mdy.univie.ac.at>

Hi Greg,

On Tue, Sep 01, 2009 at 09:05:25AM -0700, Greg Kurtzer wrote:
> On Tue, Sep 1, 2009 at 8:01 AM, Stefan Boresch<stefan at mdy.univie.ac.at> wrote:
> Yes, we know. We were planning on releasing the live media installer
> already, but then there were several high priority exploits that were
> found in the current kernels.
> 
> For testing purposes, they can be found at:
> http://altruistic.infiscale.org/isos. Please expect a new one in the
> next day or so, then we will push that to the live mirrors.

OK, I thought something like this ...

> > * Why is the preshell option not documented anymore. I knew about it
> > because I needed it in the past to manually load an ethernet driver;
> > it was docemnted then. This time I had to read the init script to
> > remember the exact "incantation"
> 
> It was documented at one point in the wiki, but I am not sure what
> happened to it or if someone removed it.

I know; that's where I found the option some months ago when it was
*very* helpful as at the time the installer made a wrong choice in
ethernet driver ... At least I knew that something most likely
existed. My general impression of the wiki is that some editors seem
to be quick to delete stuff once newer (= better?) versions appear.
E.g., the one liner outlining which modules to install for nvidia also
were online for a few months; a few weeks ago when I wanted to look at
them for a new machine they were gone and could only be "resurrected"
by google (meaning the page is still available, but there is no
direct link pointing to it)

[snip]

> Since we are moving the installer to the live media solution I don't
> think most of these issues will continue to be a problem. If you have
> the opportunity, I would be very interested in some feedback on the
> new live media installer.
> 

that makes sense. I will certainly test the new installer on new machines;
unfortunately, those needing the jmicron module are already in production (my
co-workers wouldn't appreciate if I told them to move those 12TB (or so)
somewhere else for a quick test ;-)

Speaking of installers; any chance for a net-install, or maybe a "howto"
how one could install over the net? Until caos, it was some years that I
actually inserted a CD somewhere ...
[wiki.caoslinux.org/Caos_NSA_1_Installation_-_Soekris_net5501 describes
an install that relies in part on a netboot]

Best regards,

Stefan

-- 
Stefan Boresch
Institute for Computational Biological Chemistry
University of Vienna, Waehringerstr. 17       A-1090 Vienna, Austria
Phone: -43-1-427752715                        Fax:   -43-1-427752790

From stefan at mdy.univie.ac.at  Wed Sep  2 05:17:19 2009
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 2 Sep 2009 14:17:19 +0200
Subject: [Caos] Conflict nvidia-libs and xorg-server-glx
Message-ID: <20090902121719.GE10349@loop.mdy.univie.ac.at>

[applies to stuff in nsa-testing]

This has occured to me for all nvidia-*-185.* packages for
several weeks: After 'smart update' (testing repository enabled),
'smart upgrade' fails with 

Committing transaction...
Preparing...                    ######################################## [  0%]
error: file /usr/lib64/xorg/modules/extensions/libglx.so from install of nvidia-libs-185.18.14-16.caos.x86_64 conflicts with file from package xorg-server-glx-1.4-8.caos.x86_64

I am working around that by forcing install of the nvidia packages by
hand; then doing 'smart upgrade' for the rest, but nice it is not ...

[I assume that I want libglx.so from the nvidia.rpm; things seem to work
that way]

Cheers,

Stefan

-- 
Stefan Boresch
Institute for Computational Biological Chemistry
University of Vienna, Waehringerstr. 17       A-1090 Vienna, Austria
Phone: -43-1-427752715                        Fax:   -43-1-427752790

From gmkurtzer at gmail.com  Wed Sep  2 07:50:02 2009
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 2 Sep 2009 07:50:02 -0700
Subject: [Caos] Conflict nvidia-libs and xorg-server-glx
In-Reply-To: <20090902121719.GE10349@loop.mdy.univie.ac.at>
References: <20090902121719.GE10349@loop.mdy.univie.ac.at>
Message-ID: <571f1a060909020750u4aebd406x4e203b519f9a372@mail.gmail.com>

Hi Stefan,

There is an update available in nsa-testing now that adds a Conflicts
tag for xorg-server-glx. Can you give another upgrade a test and let
me know how it goes?

Thanks!
Greg


On Wed, Sep 2, 2009 at 5:17 AM, Stefan Boresch<stefan at mdy.univie.ac.at> wrote:
> [applies to stuff in nsa-testing]
>
> This has occured to me for all nvidia-*-185.* packages for
> several weeks: After 'smart update' (testing repository enabled),
> 'smart upgrade' fails with
>
> Committing transaction...
> Preparing... ? ? ? ? ? ? ? ? ? ?######################################## [ ?0%]
> error: file /usr/lib64/xorg/modules/extensions/libglx.so from install of nvidia-libs-185.18.14-16.caos.x86_64 conflicts with file from package xorg-server-glx-1.4-8.caos.x86_64
>
> I am working around that by forcing install of the nvidia packages by
> hand; then doing 'smart upgrade' for the rest, but nice it is not ...
>
> [I assume that I want libglx.so from the nvidia.rpm; things seem to work
> that way]
>
> Cheers,
>
> Stefan
>
> --
> Stefan Boresch
> Institute for Computational Biological Chemistry
> University of Vienna, Waehringerstr. 17 ? ? ? A-1090 Vienna, Austria
> Phone: -43-1-427752715 ? ? ? ? ? ? ? ? ? ? ? ?Fax: ? -43-1-427752790
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
>



-- 
Greg Kurtzer
http://www.infiscale.com/
http://www.perceus.org/
http://www.caoslinux.org/

From stefan at mdy.univie.ac.at  Thu Sep  3 01:50:18 2009
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Thu, 3 Sep 2009 10:50:18 +0200
Subject: [Caos] Conflict nvidia-libs and xorg-server-glx
In-Reply-To: <571f1a060909020750u4aebd406x4e203b519f9a372@mail.gmail.com>
References: <20090902121719.GE10349@loop.mdy.univie.ac.at>
	<571f1a060909020750u4aebd406x4e203b519f9a372@mail.gmail.com>
Message-ID: <20090903085018.GF10349@loop.mdy.univie.ac.at>

Hi Greg,

On Wed, Sep 02, 2009 at 07:50:02AM -0700, Greg Kurtzer wrote:
> Hi Stefan,
> 
> There is an update available in nsa-testing now that adds a Conflicts
> tag for xorg-server-glx. Can you give another upgrade a test and let
> me know how it goes?
> 

Just tried, and it did this

-----------------------------------------------------
Computing transaction...

Upgrading packages (7):
  nvidia-cuda-185.18.14-17.caos at x86_64
  nvidia-cuda-devel-185.18.14-17.caos at x86_64
  nvidia-devel-185.18.14-17.caos at x86_64
  nvidia-kmods-185.18.14-17.caos at x86_64
  nvidia-libs-185.18.14-17.caos at x86_64
  nvidia-utils-185.18.14-17.caos at x86_64
  perl-5.10.0-13.caos at x86_64

Removing packages (2):
  metapkg-gl-1.2-162.nsa1 at x86_64         xorg-server-glx-1.4-8.caos at x86_64

29.7MB of package files are needed. 508.5kB will be freed.
-----------------------------------------------------

Is this what you intended? (Doesn't xorg-server-glx contain some other
stuff that might / may be needed for GL related stuff ??)

I use the chance to report another little snag that I see constantly
on this machine ... (output from the same smart upgrade run)

-----------------------------------------------------
   4:Installing nvidia-kmods    ######################################## [ 25%]
Output from nvidia-kmods-185.18.14-17.caos at x86_64:
WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-amd.ko ignored, due to loop
WARNING: Loop detected: /lib/modules/2.6.30.4-3.caos/updates/kvm.ko which needs kvm.ko again!
WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm.ko ignored, due to loopWARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-intel.ko ignored, due to loop
   5:Cleaning nvidia-kmods      ######################################## [ 31%]
WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-amd.ko ignored, due to loop
WARNING: Loop detected: /lib/modules/2.6.30.4-3.caos/updates/kvm.ko which needs kvm.ko again!
WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm.ko ignored, due to loopWARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-intel.ko ignored, due to loop
   6:Removing xorg-server-glx   ######################################## [ 37%]
-----------------------------------------------------

Since I don't use kvm, I don't know whether this is harmless or less so ...

Thanks,

Stefan

-- 
Stefan Boresch
Institute for Computational Biological Chemistry
University of Vienna, Waehringerstr. 17       A-1090 Vienna, Austria
Phone: -43-1-427752715                        Fax:   -43-1-427752790

From gmkurtzer at gmail.com  Thu Sep  3 10:37:31 2009
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Thu, 3 Sep 2009 10:37:31 -0700
Subject: [Caos] Conflict nvidia-libs and xorg-server-glx
In-Reply-To: <20090903085018.GF10349@loop.mdy.univie.ac.at>
References: <20090902121719.GE10349@loop.mdy.univie.ac.at>
	<571f1a060909020750u4aebd406x4e203b519f9a372@mail.gmail.com>
	<20090903085018.GF10349@loop.mdy.univie.ac.at>
Message-ID: <571f1a060909031037g7ddd85b8ndc1650dfeb16a799@mail.gmail.com>

Hi Stefan,

The smart output of removing the xorg-server-glx is perfect as it only
provides the conflicting library (that is why it was separated into a
sub-package).

The install warnings about kvm-kmods isn't harmful, but it isn't
helpful either. It is because KVM provides kernel modules with KVM,
and so does our kernel. So there is a duplicate set of kernel modules
for kvm*.ko. Chances are we are going to remove the kvm-kmods package
and just use the KVM support that is in the kernel.

Thanks,
Greg

On Thu, Sep 3, 2009 at 1:50 AM, Stefan Boresch<stefan at mdy.univie.ac.at> wrote:
> Hi Greg,
>
> On Wed, Sep 02, 2009 at 07:50:02AM -0700, Greg Kurtzer wrote:
>> Hi Stefan,
>>
>> There is an update available in nsa-testing now that adds a Conflicts
>> tag for xorg-server-glx. Can you give another upgrade a test and let
>> me know how it goes?
>>
>
> Just tried, and it did this
>
> -----------------------------------------------------
> Computing transaction...
>
> Upgrading packages (7):
> ?nvidia-cuda-185.18.14-17.caos at x86_64
> ?nvidia-cuda-devel-185.18.14-17.caos at x86_64
> ?nvidia-devel-185.18.14-17.caos at x86_64
> ?nvidia-kmods-185.18.14-17.caos at x86_64
> ?nvidia-libs-185.18.14-17.caos at x86_64
> ?nvidia-utils-185.18.14-17.caos at x86_64
> ?perl-5.10.0-13.caos at x86_64
>
> Removing packages (2):
> ?metapkg-gl-1.2-162.nsa1 at x86_64 ? ? ? ? xorg-server-glx-1.4-8.caos at x86_64
>
> 29.7MB of package files are needed. 508.5kB will be freed.
> -----------------------------------------------------
>
> Is this what you intended? (Doesn't xorg-server-glx contain some other
> stuff that might / may be needed for GL related stuff ??)
>
> I use the chance to report another little snag that I see constantly
> on this machine ... (output from the same smart upgrade run)
>
> -----------------------------------------------------
> ? 4:Installing nvidia-kmods ? ?######################################## [ 25%]
> Output from nvidia-kmods-185.18.14-17.caos at x86_64:
> WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-amd.ko ignored, due to loop
> WARNING: Loop detected: /lib/modules/2.6.30.4-3.caos/updates/kvm.ko which needs kvm.ko again!
> WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm.ko ignored, due to loopWARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-intel.ko ignored, due to loop
> ? 5:Cleaning nvidia-kmods ? ? ?######################################## [ 31%]
> WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-amd.ko ignored, due to loop
> WARNING: Loop detected: /lib/modules/2.6.30.4-3.caos/updates/kvm.ko which needs kvm.ko again!
> WARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm.ko ignored, due to loopWARNING: Module /lib/modules/2.6.30.4-3.caos/updates/kvm-intel.ko ignored, due to loop
> ? 6:Removing xorg-server-glx ? ######################################## [ 37%]
> -----------------------------------------------------
>
> Since I don't use kvm, I don't know whether this is harmless or less so ...
>
> Thanks,
>
> Stefan
>
> --
> Stefan Boresch
> Institute for Computational Biological Chemistry
> University of Vienna, Waehringerstr. 17 ? ? ? A-1090 Vienna, Austria
> Phone: -43-1-427752715 ? ? ? ? ? ? ? ? ? ? ? ?Fax: ? -43-1-427752790
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
>



-- 
Greg Kurtzer
http://www.infiscale.com/
http://www.perceus.org/
http://www.caoslinux.org/

From Eliezer.Rosengaus at kla-tencor.com  Tue Sep 22 08:54:39 2009
From: Eliezer.Rosengaus at kla-tencor.com (Rosengaus, Eliezer)
Date: Tue, 22 Sep 2009 08:54:39 -0700
Subject: [Caos] Shutdown problems
Message-ID: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com>

I have a cluster of boxes running caos linux (provisioned by perceus).
Each box is a dual-socket motherboard (Supermicro X8DTN). When I try to
shut down a node by issuing "shutdown -f -h now", the shutdown proceeds
down to a point where one of the CPUs is shut down, but the system then
freezes and never completes the powerdown sequence. Has anyone seen this
before or have any idea of how to fix it?

Eliezer Rosengaus

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.infiscale.org/pipermail/caos/attachments/20090922/86f573a6/attachment.html 

From gmk at infiscale.org  Tue Sep 22 10:10:52 2009
From: gmk at infiscale.org (Greg Kurtzer)
Date: Tue, 22 Sep 2009 10:10:52 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com>
References: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com>
Message-ID: <571f1a060909221010l799d7c5dk5d49a17f93372dee@mail.gmail.com>

Hi

What are the last things printed to the screen?


On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
<Eliezer.Rosengaus at kla-tencor.com> wrote:
> I have a cluster of boxes running caos linux (provisioned by perceus). Each
> box is a dual-socket motherboard (Supermicro X8DTN). When I try to shut down
> a node by issuing "shutdown -f -h now", the shutdown proceeds down to a
> point where one of the CPUs is shut down, but the system then freezes and
> never completes the powerdown sequence. Has anyone seen this before or have
> any idea of how to fix it?
>
> Eliezer Rosengaus
>
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
>
>



-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com

From Eliezer.Rosengaus at kla-tencor.com  Tue Sep 22 10:30:40 2009
From: Eliezer.Rosengaus at kla-tencor.com (Rosengaus, Eliezer)
Date: Tue, 22 Sep 2009 10:30:40 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <571f1a060909221010l799d7c5dk5d49a17f93372dee@mail.gmail.com>
References: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com>
	<571f1a060909221010l799d7c5dk5d49a17f93372dee@mail.gmail.com>
Message-ID: <87B5F8D7EA676B4EA235361F2250E6650281F343@CA1EXCLV06.adcorp.kla-tencor.com>

Disabling non-boot CPUs
...
CPU 1 is now offline






SYSTEM FREEZES 


Eliezer Rosengaus
Sr. Systems Architect, Image Computer
KLA-Tencor Corp.
5 Technology Dr.
MS 5-1245
Milpitas, CA 95035

(408)875-5196


-----Original Message-----
From: caos-bounces at lists.infiscale.org
[mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
Sent: Tuesday, September 22, 2009 10:11 AM
To: Caos Linux discussion and support forum
Subject: Re: [Caos] Shutdown problems

Hi

What are the last things printed to the screen?


On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
<Eliezer.Rosengaus at kla-tencor.com> wrote:
> I have a cluster of boxes running caos linux (provisioned by perceus).
Each
> box is a dual-socket motherboard (Supermicro X8DTN). When I try to
shut down
> a node by issuing "shutdown -f -h now", the shutdown proceeds down to
a
> point where one of the CPUs is shut down, but the system then freezes
and
> never completes the powerdown sequence. Has anyone seen this before or
have
> any idea of how to fix it?
>
> Eliezer Rosengaus
>
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
>
>



-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com
_______________________________________________
Caos mailing list
Caos at lists.infiscale.org
http://lists.infiscale.org/mailman/listinfo/caos

From gmk at infiscale.org  Tue Sep 22 11:02:12 2009
From: gmk at infiscale.org (Greg Kurtzer)
Date: Tue, 22 Sep 2009 11:02:12 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <87B5F8D7EA676B4EA235361F2250E6650281F343@CA1EXCLV06.adcorp.kla-tencor.com>
References: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com>
	<571f1a060909221010l799d7c5dk5d49a17f93372dee@mail.gmail.com>
	<87B5F8D7EA676B4EA235361F2250E6650281F343@CA1EXCLV06.adcorp.kla-tencor.com>
Message-ID: <20090922180212.GF15187@infiscale.org>

Which kernel are you running in that VNFS?



On Tuesday, 22 September 2009, at 10:30:40 (-0700),
Rosengaus, Eliezer wrote:

> Disabling non-boot CPUs
> ...
> CPU 1 is now offline
> 
> 
> 
> 
> 
> 
> SYSTEM FREEZES 
> 
> 
> Eliezer Rosengaus
> Sr. Systems Architect, Image Computer
> KLA-Tencor Corp.
> 5 Technology Dr.
> MS 5-1245
> Milpitas, CA 95035
> 
> (408)875-5196
> 
> 
> -----Original Message-----
> From: caos-bounces at lists.infiscale.org
> [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> Sent: Tuesday, September 22, 2009 10:11 AM
> To: Caos Linux discussion and support forum
> Subject: Re: [Caos] Shutdown problems
> 
> Hi
> 
> What are the last things printed to the screen?
> 
> 
> On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
> <Eliezer.Rosengaus at kla-tencor.com> wrote:
> > I have a cluster of boxes running caos linux (provisioned by perceus).
> Each
> > box is a dual-socket motherboard (Supermicro X8DTN). When I try to
> shut down
> > a node by issuing "shutdown -f -h now", the shutdown proceeds down to
> a
> > point where one of the CPUs is shut down, but the system then freezes
> and
> > never completes the powerdown sequence. Has anyone seen this before or
> have
> > any idea of how to fix it?
> >
> > Eliezer Rosengaus
> >
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> >
> >
> 
> 
> 
> -- 
> Greg M. Kurtzer
> Chief Technology Officer
> HPC Systems Architect
> Infiscale, Inc. - http://www.infiscale.com
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos

-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com

From Eliezer.Rosengaus at kla-tencor.com  Tue Sep 22 11:35:46 2009
From: Eliezer.Rosengaus at kla-tencor.com (Rosengaus, Eliezer)
Date: Tue, 22 Sep 2009 11:35:46 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <20090922180212.GF15187@infiscale.org>
References: <87B5F8D7EA676B4EA235361F2250E6650281F2F1@CA1EXCLV06.adcorp.kla-tencor.com><571f1a060909221010l799d7c5dk5d49a17f93372dee@mail.gmail.com><87B5F8D7EA676B4EA235361F2250E6650281F343@CA1EXCLV06.adcorp.kla-tencor.com>
	<20090922180212.GF15187@infiscale.org>
Message-ID: <87B5F8D7EA676B4EA235361F2250E6650281F38E@CA1EXCLV06.adcorp.kla-tencor.com>

the caos vnfs on the perceus site:

Linux node001 2.6.28.3-1.nsa1 #1 SMP Wed Feb 4 23:08:20 PST 2009 x86_64
x86_64 x86_64 GNU/Linux


Eliezer Rosengaus
Sr. Systems Architect, Image Computer
KLA-Tencor Corp.
5 Technology Dr.
MS 5-1245
Milpitas, CA 95035

(408)875-5196


-----Original Message-----
From: caos-bounces at lists.infiscale.org
[mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
Sent: Tuesday, September 22, 2009 11:02 AM
To: Caos Linux discussion and support forum
Subject: Re: [Caos] Shutdown problems

Which kernel are you running in that VNFS?



On Tuesday, 22 September 2009, at 10:30:40 (-0700),
Rosengaus, Eliezer wrote:

> Disabling non-boot CPUs
> ...
> CPU 1 is now offline
> 
> 
> 
> 
> 
> 
> SYSTEM FREEZES 
> 
> 
> Eliezer Rosengaus
> Sr. Systems Architect, Image Computer
> KLA-Tencor Corp.
> 5 Technology Dr.
> MS 5-1245
> Milpitas, CA 95035
> 
> (408)875-5196
> 
> 
> -----Original Message-----
> From: caos-bounces at lists.infiscale.org
> [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> Sent: Tuesday, September 22, 2009 10:11 AM
> To: Caos Linux discussion and support forum
> Subject: Re: [Caos] Shutdown problems
> 
> Hi
> 
> What are the last things printed to the screen?
> 
> 
> On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
> <Eliezer.Rosengaus at kla-tencor.com> wrote:
> > I have a cluster of boxes running caos linux (provisioned by
perceus).
> Each
> > box is a dual-socket motherboard (Supermicro X8DTN). When I try to
> shut down
> > a node by issuing "shutdown -f -h now", the shutdown proceeds down
to
> a
> > point where one of the CPUs is shut down, but the system then
freezes
> and
> > never completes the powerdown sequence. Has anyone seen this before
or
> have
> > any idea of how to fix it?
> >
> > Eliezer Rosengaus
> >
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> >
> >
> 
> 
> 
> -- 
> Greg M. Kurtzer
> Chief Technology Officer
> HPC Systems Architect
> Infiscale, Inc. - http://www.infiscale.com
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos

-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com
_______________________________________________
Caos mailing list
Caos at lists.infiscale.org
http://lists.infiscale.org/mailman/listinfo/caos

From gmk at infiscale.org  Tue Sep 22 12:10:13 2009
From: gmk at infiscale.org (Greg Kurtzer)
Date: Tue, 22 Sep 2009 12:10:13 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <87B5F8D7EA676B4EA235361F2250E6650281F38E@CA1EXCLV06.adcorp.kla-tencor.com>
References: <20090922180212.GF15187@infiscale.org>
	<87B5F8D7EA676B4EA235361F2250E6650281F38E@CA1EXCLV06.adcorp.kla-tencor.com>
Message-ID: <20090922191013.GG15187@infiscale.org>

There is an updated kernel on the mirrors. Give that a try and let us know.

On Tuesday, 22 September 2009, at 11:35:46 (-0700),
Rosengaus, Eliezer wrote:

> the caos vnfs on the perceus site:
> 
> Linux node001 2.6.28.3-1.nsa1 #1 SMP Wed Feb 4 23:08:20 PST 2009 x86_64
> x86_64 x86_64 GNU/Linux
> 
> 
> Eliezer Rosengaus
> Sr. Systems Architect, Image Computer
> KLA-Tencor Corp.
> 5 Technology Dr.
> MS 5-1245
> Milpitas, CA 95035
> 
> (408)875-5196
> 
> 
> -----Original Message-----
> From: caos-bounces at lists.infiscale.org
> [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> Sent: Tuesday, September 22, 2009 11:02 AM
> To: Caos Linux discussion and support forum
> Subject: Re: [Caos] Shutdown problems
> 
> Which kernel are you running in that VNFS?
> 
> 
> 
> On Tuesday, 22 September 2009, at 10:30:40 (-0700),
> Rosengaus, Eliezer wrote:
> 
> > Disabling non-boot CPUs
> > ...
> > CPU 1 is now offline
> > 
> > 
> > 
> > 
> > 
> > 
> > SYSTEM FREEZES 
> > 
> > 
> > Eliezer Rosengaus
> > Sr. Systems Architect, Image Computer
> > KLA-Tencor Corp.
> > 5 Technology Dr.
> > MS 5-1245
> > Milpitas, CA 95035
> > 
> > (408)875-5196
> > 
> > 
> > -----Original Message-----
> > From: caos-bounces at lists.infiscale.org
> > [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> > Sent: Tuesday, September 22, 2009 10:11 AM
> > To: Caos Linux discussion and support forum
> > Subject: Re: [Caos] Shutdown problems
> > 
> > Hi
> > 
> > What are the last things printed to the screen?
> > 
> > 
> > On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
> > <Eliezer.Rosengaus at kla-tencor.com> wrote:
> > > I have a cluster of boxes running caos linux (provisioned by
> perceus).
> > Each
> > > box is a dual-socket motherboard (Supermicro X8DTN). When I try to
> > shut down
> > > a node by issuing "shutdown -f -h now", the shutdown proceeds down
> to
> > a
> > > point where one of the CPUs is shut down, but the system then
> freezes
> > and
> > > never completes the powerdown sequence. Has anyone seen this before
> or
> > have
> > > any idea of how to fix it?
> > >
> > > Eliezer Rosengaus
> > >
> > > _______________________________________________
> > > Caos mailing list
> > > Caos at lists.infiscale.org
> > > http://lists.infiscale.org/mailman/listinfo/caos
> > >
> > >
> > 
> > 
> > 
> > -- 
> > Greg M. Kurtzer
> > Chief Technology Officer
> > HPC Systems Architect
> > Infiscale, Inc. - http://www.infiscale.com
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> 
> -- 
> Greg M. Kurtzer
> Chief Technology Officer
> HPC Systems Architect
> Infiscale, Inc. - http://www.infiscale.com
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos

-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com

From Eliezer.Rosengaus at kla-tencor.com  Tue Sep 22 15:00:14 2009
From: Eliezer.Rosengaus at kla-tencor.com (Rosengaus, Eliezer)
Date: Tue, 22 Sep 2009 15:00:14 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <20090922191013.GG15187@infiscale.org>
References: <20090922180212.GF15187@infiscale.org><87B5F8D7EA676B4EA235361F2250E6650281F38E@CA1EXCLV06.adcorp.kla-tencor.com>
	<20090922191013.GG15187@infiscale.org>
Message-ID: <87B5F8D7EA676B4EA235361F2250E6650281F471@CA1EXCLV06.adcorp.kla-tencor.com>

Sorry for the newbie questions, but I copied the new kernel files to the
vnfs /boot directory. How do I change the boot kernel specification
(like in the grub menu.lst) in perceus?
By the way, the Centos capsule (older kernel 2.6.18...) shuts down fine.



Eliezer Rosengaus
Sr. Systems Architect, Image Computer
KLA-Tencor Corp.
5 Technology Dr.
MS 5-1245
Milpitas, CA 95035

(408)875-5196


-----Original Message-----
From: caos-bounces at lists.infiscale.org
[mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
Sent: Tuesday, September 22, 2009 12:10 PM
To: Caos Linux discussion and support forum
Subject: Re: [Caos] Shutdown problems

There is an updated kernel on the mirrors. Give that a try and let us
know.

On Tuesday, 22 September 2009, at 11:35:46 (-0700),
Rosengaus, Eliezer wrote:

> the caos vnfs on the perceus site:
> 
> Linux node001 2.6.28.3-1.nsa1 #1 SMP Wed Feb 4 23:08:20 PST 2009
x86_64
> x86_64 x86_64 GNU/Linux
> 
> 
> Eliezer Rosengaus
> Sr. Systems Architect, Image Computer
> KLA-Tencor Corp.
> 5 Technology Dr.
> MS 5-1245
> Milpitas, CA 95035
> 
> (408)875-5196
> 
> 
> -----Original Message-----
> From: caos-bounces at lists.infiscale.org
> [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> Sent: Tuesday, September 22, 2009 11:02 AM
> To: Caos Linux discussion and support forum
> Subject: Re: [Caos] Shutdown problems
> 
> Which kernel are you running in that VNFS?
> 
> 
> 
> On Tuesday, 22 September 2009, at 10:30:40 (-0700),
> Rosengaus, Eliezer wrote:
> 
> > Disabling non-boot CPUs
> > ...
> > CPU 1 is now offline
> > 
> > 
> > 
> > 
> > 
> > 
> > SYSTEM FREEZES 
> > 
> > 
> > Eliezer Rosengaus
> > Sr. Systems Architect, Image Computer
> > KLA-Tencor Corp.
> > 5 Technology Dr.
> > MS 5-1245
> > Milpitas, CA 95035
> > 
> > (408)875-5196
> > 
> > 
> > -----Original Message-----
> > From: caos-bounces at lists.infiscale.org
> > [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> > Sent: Tuesday, September 22, 2009 10:11 AM
> > To: Caos Linux discussion and support forum
> > Subject: Re: [Caos] Shutdown problems
> > 
> > Hi
> > 
> > What are the last things printed to the screen?
> > 
> > 
> > On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
> > <Eliezer.Rosengaus at kla-tencor.com> wrote:
> > > I have a cluster of boxes running caos linux (provisioned by
> perceus).
> > Each
> > > box is a dual-socket motherboard (Supermicro X8DTN). When I try to
> > shut down
> > > a node by issuing "shutdown -f -h now", the shutdown proceeds down
> to
> > a
> > > point where one of the CPUs is shut down, but the system then
> freezes
> > and
> > > never completes the powerdown sequence. Has anyone seen this
before
> or
> > have
> > > any idea of how to fix it?
> > >
> > > Eliezer Rosengaus
> > >
> > > _______________________________________________
> > > Caos mailing list
> > > Caos at lists.infiscale.org
> > > http://lists.infiscale.org/mailman/listinfo/caos
> > >
> > >
> > 
> > 
> > 
> > -- 
> > Greg M. Kurtzer
> > Chief Technology Officer
> > HPC Systems Architect
> > Infiscale, Inc. - http://www.infiscale.com
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> 
> -- 
> Greg M. Kurtzer
> Chief Technology Officer
> HPC Systems Architect
> Infiscale, Inc. - http://www.infiscale.com
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos

-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com
_______________________________________________
Caos mailing list
Caos at lists.infiscale.org
http://lists.infiscale.org/mailman/listinfo/caos

From gmk at infiscale.org  Tue Sep 22 16:38:27 2009
From: gmk at infiscale.org (Greg Kurtzer)
Date: Tue, 22 Sep 2009 16:38:27 -0700
Subject: [Caos] Shutdown problems
In-Reply-To: <87B5F8D7EA676B4EA235361F2250E6650281F471@CA1EXCLV06.adcorp.kla-tencor.com>
References: <20090922191013.GG15187@infiscale.org>
	<87B5F8D7EA676B4EA235361F2250E6650281F471@CA1EXCLV06.adcorp.kla-tencor.com>
Message-ID: <20090922233827.GH15187@infiscale.org>

Give this a try:

# perceus vnfs mount [vnfs name]
# smart -o rpm-root=/mnt/[vnfs name] upgrade caos-kernel
# perceus vnfs umount [vnfs name]
# vi /etc/perceus/vnfs/[vnfs name]/config

Define the kernel you wish to boot.

Reboot the node.

Greg



On Tuesday, 22 September 2009, at 15:00:14 (-0700),
Rosengaus, Eliezer wrote:

> Sorry for the newbie questions, but I copied the new kernel files to the
> vnfs /boot directory. How do I change the boot kernel specification
> (like in the grub menu.lst) in perceus?
> By the way, the Centos capsule (older kernel 2.6.18...) shuts down fine.
> 
> 
> 
> Eliezer Rosengaus
> Sr. Systems Architect, Image Computer
> KLA-Tencor Corp.
> 5 Technology Dr.
> MS 5-1245
> Milpitas, CA 95035
> 
> (408)875-5196
> 
> 
> -----Original Message-----
> From: caos-bounces at lists.infiscale.org
> [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> Sent: Tuesday, September 22, 2009 12:10 PM
> To: Caos Linux discussion and support forum
> Subject: Re: [Caos] Shutdown problems
> 
> There is an updated kernel on the mirrors. Give that a try and let us
> know.
> 
> On Tuesday, 22 September 2009, at 11:35:46 (-0700),
> Rosengaus, Eliezer wrote:
> 
> > the caos vnfs on the perceus site:
> > 
> > Linux node001 2.6.28.3-1.nsa1 #1 SMP Wed Feb 4 23:08:20 PST 2009
> x86_64
> > x86_64 x86_64 GNU/Linux
> > 
> > 
> > Eliezer Rosengaus
> > Sr. Systems Architect, Image Computer
> > KLA-Tencor Corp.
> > 5 Technology Dr.
> > MS 5-1245
> > Milpitas, CA 95035
> > 
> > (408)875-5196
> > 
> > 
> > -----Original Message-----
> > From: caos-bounces at lists.infiscale.org
> > [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> > Sent: Tuesday, September 22, 2009 11:02 AM
> > To: Caos Linux discussion and support forum
> > Subject: Re: [Caos] Shutdown problems
> > 
> > Which kernel are you running in that VNFS?
> > 
> > 
> > 
> > On Tuesday, 22 September 2009, at 10:30:40 (-0700),
> > Rosengaus, Eliezer wrote:
> > 
> > > Disabling non-boot CPUs
> > > ...
> > > CPU 1 is now offline
> > > 
> > > 
> > > 
> > > 
> > > 
> > > 
> > > SYSTEM FREEZES 
> > > 
> > > 
> > > Eliezer Rosengaus
> > > Sr. Systems Architect, Image Computer
> > > KLA-Tencor Corp.
> > > 5 Technology Dr.
> > > MS 5-1245
> > > Milpitas, CA 95035
> > > 
> > > (408)875-5196
> > > 
> > > 
> > > -----Original Message-----
> > > From: caos-bounces at lists.infiscale.org
> > > [mailto:caos-bounces at lists.infiscale.org] On Behalf Of Greg Kurtzer
> > > Sent: Tuesday, September 22, 2009 10:11 AM
> > > To: Caos Linux discussion and support forum
> > > Subject: Re: [Caos] Shutdown problems
> > > 
> > > Hi
> > > 
> > > What are the last things printed to the screen?
> > > 
> > > 
> > > On Tue, Sep 22, 2009 at 8:54 AM, Rosengaus, Eliezer
> > > <Eliezer.Rosengaus at kla-tencor.com> wrote:
> > > > I have a cluster of boxes running caos linux (provisioned by
> > perceus).
> > > Each
> > > > box is a dual-socket motherboard (Supermicro X8DTN). When I try to
> > > shut down
> > > > a node by issuing "shutdown -f -h now", the shutdown proceeds down
> > to
> > > a
> > > > point where one of the CPUs is shut down, but the system then
> > freezes
> > > and
> > > > never completes the powerdown sequence. Has anyone seen this
> before
> > or
> > > have
> > > > any idea of how to fix it?
> > > >
> > > > Eliezer Rosengaus
> > > >
> > > > _______________________________________________
> > > > Caos mailing list
> > > > Caos at lists.infiscale.org
> > > > http://lists.infiscale.org/mailman/listinfo/caos
> > > >
> > > >
> > > 
> > > 
> > > 
> > > -- 
> > > Greg M. Kurtzer
> > > Chief Technology Officer
> > > HPC Systems Architect
> > > Infiscale, Inc. - http://www.infiscale.com
> > > _______________________________________________
> > > Caos mailing list
> > > Caos at lists.infiscale.org
> > > http://lists.infiscale.org/mailman/listinfo/caos
> > > _______________________________________________
> > > Caos mailing list
> > > Caos at lists.infiscale.org
> > > http://lists.infiscale.org/mailman/listinfo/caos
> > 
> > -- 
> > Greg M. Kurtzer
> > Chief Technology Officer
> > HPC Systems Architect
> > Infiscale, Inc. - http://www.infiscale.com
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> > _______________________________________________
> > Caos mailing list
> > Caos at lists.infiscale.org
> > http://lists.infiscale.org/mailman/listinfo/caos
> 
> -- 
> Greg M. Kurtzer
> Chief Technology Officer
> HPC Systems Architect
> Infiscale, Inc. - http://www.infiscale.com
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos
> _______________________________________________
> Caos mailing list
> Caos at lists.infiscale.org
> http://lists.infiscale.org/mailman/listinfo/caos

-- 
Greg M. Kurtzer
Chief Technology Officer
HPC Systems Architect
Infiscale, Inc. - http://www.infiscale.com