From jsalbinson at arcastro.co.uk  Thu May  1 11:53:04 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Thu, 1 May 2008 19:53:04 +0100
Subject: [Warewulf] Perceus 1.3.7 - no response to node dhcp requests.
In-Reply-To: <571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<200804302052.29161.jsalbinson@arcastro.co.uk>
	<571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>
Message-ID: <200805011953.05480.jsalbinson@arcastro.co.uk>

Hi Greg,
Only the perceus-dnsmasq startup messages (now lost - see below).
No DNSREQUEST or DNSOFFER messages.
Then I disabled the firewall (sigh), and saw a DNSREQ and a DNSOFFer, but no 
moreDNS mesages and no take up of pxelinux.0
I tried installing drbl to see if the network was OK - it was, but the kernel 
on the athlon host disagreed with the duron client.
AS I had also tried to compile OSCAR as well, earlier, the system was such a 
mess I lost contact with my home network...

I am now doing a clean install of Scientific Linux 5.1 (again), and will try a 
clean install of perceus 1.3.7 alone. No warewulf of any order, and no other 
software.
I'll be back...

Thanks, JAmes ALbinson

On Thursday 01 May 2008, Greg Kurtzer wrote:
> Are there any syslog messages from perceus-dnsmasq on the master?
>
>
>
> On Wed, Apr 30, 2008 at 12:52 PM, James Albinson
>
> <jsalbinson at arcastro.co.uk> wrote:
> > Another scenario...
> >  The situation.
> >
> >  I have a test clusterhost, Athlon 1GHz, 768Mb, 20Gb HD.
> >  I have installed Scientific Linux 5.1 (the latest DVD) i386 cpu version.
> >  I have d/l perceus 1.3.7 source and built from scratch, installing the
> >  dependencies as configure burped...  It all built OK in the end.
> >  Make install went OK, I ran perceus init all, and it seemed OK.
> >  I did /etc/init.d/perceus restart a few times - Seems OK. Also
> > provisiond restart. OK.
> >
> >  Then I put the pxe floppy in the client node - connnected to a switch,
> > and then to the cluster host.  I used wireshark to monitor the eth1 on
> > the clusterhost, and the node is there knocking on the door.
> >
> >  But it doesn't register with perceus-dnsmasq - the lease file is non
> >  existent /empty.
> >
> >  I went through the make install, perceus init all, and restart daemons
> > twice. Still no joy.
> >
> >  Can anyone shed some light on this? I am deeply puzzled. What other
> >  tests/monitor can I run?
> >
> >  Cheers and thanks,
> >  James Albinson
> >
> >  _______________________________________________
> >  Warewulf mailing list
> >  Warewulf at caoslinux.org
> >  http://lists.caosity.org/mailman/listinfo/warewulf




From jeff.blasius at yale.edu  Thu May  1 14:04:48 2008
From: jeff.blasius at yale.edu (Jeff Blasius)
Date: Thu, 1 May 2008 17:04:48 -0400
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL 5.1
Message-ID: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>

I'm having some problems with the tg3 driver in the initramfs. The
basic problem is that no link can be detected on the interface.

The first ifconfig up does not show any signs of negotiating a link.
If I cycle the interface down and up, using ifconfig in the init
script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
shows it to be down. If I set the ethinfo -l line in init to true,
udhcpc still can not get an address... so there probably isn't a link
even though dmesg says so.

Has anyone come across this problem before especially with the tg3
driver? Of course a RHEL install on the same hardware works fine.

Thank You,
                        jeff

-- 
Jeff Blasius / jeff.blasius at yale.edu
Phone: (203)432-9940 51 Prospect Rm. 011
High Performance Computing (HPC)
UNIX Systems Administrator, Linux Systems Design & Support (LSDS)
Yale University Information Technology Services (ITS)


From jsalbinson at arcastro.co.uk  Thu May  1 14:13:11 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Thu, 1 May 2008 22:13:11 +0100
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <200805011953.05480.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>
	<200805011953.05480.jsalbinson@arcastro.co.uk>
Message-ID: <200805012213.11744.jsalbinson@arcastro.co.uk>

Hi,
I notice that when building the SL 5.1 vfns image, some 30+Mb of Red Hat Logos 
are downloaded.
Is this necessary? Out of 100Mb of download, its an awefully big chunk. 
Maybe, if neeeded for copyright reasons, something really skinny could be 
substituted?

Regards, James Albinson


From jlforrest at berkeley.edu  Thu May  1 14:17:14 2008
From: jlforrest at berkeley.edu (Jon Forrest)
Date: Thu, 01 May 2008 14:17:14 -0700
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL
 5.1
In-Reply-To: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
Message-ID: <481A335A.1090203@berkeley.edu>

Jeff Blasius wrote:
> I'm having some problems with the tg3 driver in the initramfs. The
> basic problem is that no link can be detected on the interface.
[...]
> Has anyone come across this problem before especially with the tg3
> driver? Of course a RHEL install on the same hardware works fine.

I might have seen something like this. As I recall, it takes
several seconds for the link to come up. I had to put a sleep
statement after the ifconfig command so that the following
commands that did things over the network would work.

It might have something to do with auto negotiation but
I'm not sure.

Cordially,
-- 
Jon Forrest
Research Computing Support
College of Chemistry
173 Tan Hall
University of California Berkeley
Berkeley, CA
94720-1460
510-643-1032
jlforrest at berkeley.edu


From Darin.Perusich at cognigencorp.com  Thu May  1 14:21:39 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Thu, 01 May 2008 17:21:39 -0400
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <200805012213.11744.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>	<571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>	<200805011953.05480.jsalbinson@arcastro.co.uk>
	<200805012213.11744.jsalbinson@arcastro.co.uk>
Message-ID: <481A3463.7040803@cognigencorp.com>

You can safely remove them, all they're doing is bloating the image.

James Albinson wrote:
> Hi,
> I notice that when building the SL 5.1 vfns image, some 30+Mb of Red Hat Logos 
> are downloaded.
> Is this necessary? Out of 100Mb of download, its an awefully big chunk. 
> Maybe, if neeeded for copyright reasons, something really skinny could be 
> substituted?
> 

-- 
Darin Perusich
Unix Systems Administrator
Cognigen Corporation
395 Youngs Rd.
Williamsville, NY 14221
Phone: 716-633-3463
Email: darinper at cognigencorp.com


From mej at caoslinux.org  Thu May  1 14:48:46 2008
From: mej at caoslinux.org (Michael Jennings)
Date: Thu, 1 May 2008 14:48:46 -0700
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on	RHEL
	5.1
In-Reply-To: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
Message-ID: <20080501214846.GO3941@kainx.org>

On Thursday, 01 May 2008, at 17:04:48 (-0400),
Jeff Blasius wrote:

> I'm having some problems with the tg3 driver in the initramfs. The
> basic problem is that no link can be detected on the interface.
> 
> The first ifconfig up does not show any signs of negotiating a link.
> If I cycle the interface down and up, using ifconfig in the init
> script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
> shows it to be down. If I set the ethinfo -l line in init to true,
> udhcpc still can not get an address... so there probably isn't a link
> even though dmesg says so.
> 
> Has anyone come across this problem before especially with the tg3
> driver? Of course a RHEL install on the same hardware works fine.

While I can't say for sure what's breaking, I can say that we have
dozens of machines using the tg3 driver with Perceus 1.3.7 which are
functioning properly.  So Perceus is probably not the culprit here.

Michael

-- 
Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej at kainx.org>
Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
-----------------------------------------------------------------------
 "A little learning is a dangerous thing;  Drink deep, or taste not
  the Pierian spring:  There shallow draughts intoxicate the brain,
  And drinking largely sobers us again."
                            -- Alexander Pope, "An Essay on Criticism"


From jsalbinson at arcastro.co.uk  Fri May  2 00:22:23 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Fri, 2 May 2008 08:22:23 +0100
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <481A3463.7040803@cognigencorp.com>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<200805012213.11744.jsalbinson@arcastro.co.uk>
	<481A3463.7040803@cognigencorp.com>
Message-ID: <200805020822.24352.jsalbinson@arcastro.co.uk>

Hi Darin,
Thanks for the tip, but where is the logo stuff configured?
I had a scan through the vnfs setup file, and couldn't find the pointer to the 
file that specifies the rpms  to use in the vnfs. Am I missing something? 
(Probably yes...!).

Chhers, and Thanks, JAmes Albinson

On Thursday 01 May 2008, Darin Perusich wrote:
> You can safely remove them, all they're doing is bloating the image.
>
> James Albinson wrote:
> > Hi,
> > I notice that when building the SL 5.1 vfns image, some 30+Mb of Red Hat
> > Logos are downloaded.
> > Is this necessary? Out of 100Mb of download, its an awefully big chunk.
> > Maybe, if neeeded for copyright reasons, something really skinny could be
> > substituted?




From Darin.Perusich at cognigencorp.com  Fri May  2 05:21:53 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Fri, 02 May 2008 08:21:53 -0400
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <200805020822.24352.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<200805012213.11744.jsalbinson@arcastro.co.uk>
	<481A3463.7040803@cognigencorp.com>
	<200805020822.24352.jsalbinson@arcastro.co.uk>
Message-ID: <481B0761.9050508@cognigencorp.com>

I'm not sure but if you do 'rpm -qlif /path/to/image.png' it will tell 
you which package it's associated with.

James Albinson wrote:
> Hi Darin,
> Thanks for the tip, but where is the logo stuff configured?
> I had a scan through the vnfs setup file, and couldn't find the pointer to the 
> file that specifies the rpms  to use in the vnfs. Am I missing something? 
> (Probably yes...!).
> 
> Chhers, and Thanks, JAmes Albinson
> 
> On Thursday 01 May 2008, Darin Perusich wrote:
>> You can safely remove them, all they're doing is bloating the image.
>>
>> James Albinson wrote:
>>> Hi,
>>> I notice that when building the SL 5.1 vfns image, some 30+Mb of Red Hat
>>> Logos are downloaded.
>>> Is this necessary? Out of 100Mb of download, its an awefully big chunk.
>>> Maybe, if neeeded for copyright reasons, something really skinny could be
>>> substituted?
> 
> 

-- 
Darin Perusich
Unix Systems Administrator
Cognigen Corporation
395 Youngs Rd.
Williamsville, NY 14221
Phone: 716-633-3463
Email: darinper at cognigencorp.com


From jsalbinson at arcastro.co.uk  Fri May  2 08:32:15 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Fri, 2 May 2008 16:32:15 +0100
Subject: [Warewulf] Perceus 1.3.7 - no response to node dhcp requests. A
	clean solution and a new problem...
In-Reply-To: <200805011953.05480.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>
	<200805011953.05480.jsalbinson@arcastro.co.uk>
Message-ID: <200805021632.16681.jsalbinson@arcastro.co.uk>

Hi Greg,
I'm back...
I think it was a screwed up master node...
I rebuilt a clean SL5.1 installation, and added the perceus 1.3.7 tar ball, 
plus its dependencies. Nothing else.
I untarred the ball, and did the gnu thingy (ignoring the manual as ususl...)

# ./configure && make && make install

Then I ran the share/vnfs/scientificlinux5.1genchroot.sh
followed by chroot2stateless.sh   taking defaults on all prompts.

Then I rebooted and checked the perceus-dnsmasq was running.
Then I did the perceus init all, and then perceus vnfs import 
scientificlinux51i386 thing. Checked the /etc/perceus/*.conf files and added 
the vnfs = sl51i386 entry.

Then I rebooted again, checked things over, and started the client node.
I used pxe-on-a-disk (etherboot with all the drivers v5.4.3ish) (also a pared 
down 3c905c-tpo to suit the card - same result).
The client node mac address was seen, but not provisioned.
I deleted the node and started again, and it was provisioned!
But the error message:

ramdisks not supported with general elf arguments appeared, with a reboot/drop 
into a shell option.

At this point I grabbed the opportunity to mow the lawn, so thats as far as I 
got.

herewith the last few lines in the mesages file that relate to the setup...
May  2 13:04:34 kuwbobcluster perceus-dnsmasq[2108]: DHCPDISCOVER(eth1) 
00:01:03:c3:75:ee
May  2 13:04:34 kuwbobcluster perceus-dnsmasq[2108]: DHCPOFFER(eth1) 
192.168.0.196 00:01:03:c3:75:ee
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: DHCPREQUEST(eth1) 
192.168.0.196 00:01:03:c3:75:ee
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: DHCPACK(eth1) 
192.168.0.196 00:01:03:c3:75:ee n0000
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: TFTP 
sent /usr/local/var/lib/perceus//tftp/pxelinux.0 to 192.168.0.196
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: TFTP 
sent /usr/local/var/lib/perceus//tftp/pxelinux.cfg/default to 192.168.0.196
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: TFTP 
sent /usr/local/var/lib/perceus//tftp/kernel to 192.168.0.196
May  2 13:04:36 kuwbobcluster perceus-dnsmasq[2108]: TFTP 
sent /usr/local/var/lib/perceus//tftp/initramfs.img to 192.168.0.196
May  2 13:04:42 kuwbobcluster perceus-dnsmasq[2108]: DHCPDISCOVER(eth1) 
00:01:03:c3:75:ee
May  2 13:04:42 kuwbobcluster perceus-dnsmasq[2108]: DHCPOFFER(eth1) 
192.168.0.196 00:01:03:c3:75:ee
May  2 13:04:42 kuwbobcluster perceus-dnsmasq[2108]: DHCPREQUEST(eth1) 
192.168.0.196 00:01:03:c3:75:ee
May  2 13:04:42 kuwbobcluster perceus-dnsmasq[2108]: DHCPACK(eth1) 
192.168.0.196 00:01:03:c3:75:ee n0000
May  2 13:04:42 kuwbobcluster perceusd[2112]: Provisioning 'n0000' now...
May  2 13:04:42 kuwbobcluster mountd[1977]: authenticated mount request from 
192.168.0.196:714 for /usr/local/var/lib/perceus (/usr/local/var/lib/perceus)

The hardware setup is:
masternode: Athlon 1GHz i686, 768Mb, rtl8139 cards (x4)
client node Duron 900MHz, 384Mb (3com 3c905c-tpo)

masternode has a connection on eth0 10.0.0.7 to home router >> internet.
eth1 is 192.168.0.1 to a switch, and thence to the client node.
Intersstingly, the client node is offered .196, even though I set it up as 
starting at .128, as per defaults. Odd...

Chhers all, James Albinson



From gwleong at gmail.com  Fri May  2 09:43:51 2008
From: gwleong at gmail.com (Gary Leong)
Date: Fri, 2 May 2008 12:43:51 -0400
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL
	5.1
In-Reply-To: <20080501214846.GO3941@kainx.org>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
	<20080501214846.GO3941@kainx.org>
Message-ID: <ae974e900805020943u7c8e058bvfec9b3569def2825@mail.gmail.com>

I haven't seen this problem per se, but I can tell you if the machine
is a Dell unit with tg3, the Redhat kernel and its derivatives are
known to flip the interfaces, where eth0 is seen as the second
interface.  But I'm not sure if you're having problems with stage 1
with the perceus kernel or stage 2 with the second kernel.



On Thu, May 1, 2008 at 5:48 PM, Michael Jennings <mej at caoslinux.org> wrote:
> On Thursday, 01 May 2008, at 17:04:48 (-0400),
>
> Jeff Blasius wrote:
>
>  > I'm having some problems with the tg3 driver in the initramfs. The
>  > basic problem is that no link can be detected on the interface.
>  >
>  > The first ifconfig up does not show any signs of negotiating a link.
>  > If I cycle the interface down and up, using ifconfig in the init
>  > script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
>  > shows it to be down. If I set the ethinfo -l line in init to true,
>  > udhcpc still can not get an address... so there probably isn't a link
>  > even though dmesg says so.
>  >
>  > Has anyone come across this problem before especially with the tg3
>  > driver? Of course a RHEL install on the same hardware works fine.
>
>  While I can't say for sure what's breaking, I can say that we have
>  dozens of machines using the tg3 driver with Perceus 1.3.7 which are
>  functioning properly.  So Perceus is probably not the culprit here.
>
>  Michael
>
>  --
>  Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej at kainx.org>
>  Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
>  -----------------------------------------------------------------------
>   "A little learning is a dangerous thing;  Drink deep, or taste not
>   the Pierian spring:  There shallow draughts intoxicate the brain,
>   And drinking largely sobers us again."
>                             -- Alexander Pope, "An Essay on Criticism"
>
>
> _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>


From poknam at gmail.com  Fri May  2 19:41:53 2008
From: poknam at gmail.com (PN)
Date: Sat, 3 May 2008 10:41:53 +0800
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL
	5.1
In-Reply-To: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
Message-ID: <92daa7bf0805021941o2c225655y6b094fdb65a5a544@mail.gmail.com>

i've the same problem.
for my situation, it seems that perceus can't load the tg3 module properly.
i used perceus 1.3.7 in tg3 machines and my e1000 VMs, and both not work.
so i use 1.3.6 now, at least everything work properly.

PN

2008/5/2 Jeff Blasius <jeff.blasius at yale.edu>:

> I'm having some problems with the tg3 driver in the initramfs. The
> basic problem is that no link can be detected on the interface.
>
> The first ifconfig up does not show any signs of negotiating a link.
> If I cycle the interface down and up, using ifconfig in the init
> script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
> shows it to be down. If I set the ethinfo -l line in init to true,
> udhcpc still can not get an address... so there probably isn't a link
> even though dmesg says so.
>
> Has anyone come across this problem before especially with the tg3
> driver? Of course a RHEL install on the same hardware works fine.
>
> Thank You,
>                        jeff
>
> --
> Jeff Blasius / jeff.blasius at yale.edu
> Phone: (203)432-9940 51 Prospect Rm. 011
> High Performance Computing (HPC)
> UNIX Systems Administrator, Linux Systems Design & Support (LSDS)
> Yale University Information Technology Services (ITS)
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080503/8dfeac58/attachment.html 

From poknam at gmail.com  Fri May  2 19:41:53 2008
From: poknam at gmail.com (PN)
Date: Sat, 3 May 2008 10:41:53 +0800
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL
	5.1
In-Reply-To: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
Message-ID: <92daa7bf0805021941o2c225655y6b094fdb65a5a544@mail.gmail.com>

i've the same problem.
for my situation, it seems that perceus can't load the tg3 module properly.
i used perceus 1.3.7 in tg3 machines and my e1000 VMs, and both not work.
so i use 1.3.6 now, at least everything work properly.

PN

2008/5/2 Jeff Blasius <jeff.blasius at yale.edu>:

> I'm having some problems with the tg3 driver in the initramfs. The
> basic problem is that no link can be detected on the interface.
>
> The first ifconfig up does not show any signs of negotiating a link.
> If I cycle the interface down and up, using ifconfig in the init
> script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
> shows it to be down. If I set the ethinfo -l line in init to true,
> udhcpc still can not get an address... so there probably isn't a link
> even though dmesg says so.
>
> Has anyone come across this problem before especially with the tg3
> driver? Of course a RHEL install on the same hardware works fine.
>
> Thank You,
>                        jeff
>
> --
> Jeff Blasius / jeff.blasius at yale.edu
> Phone: (203)432-9940 51 Prospect Rm. 011
> High Performance Computing (HPC)
> UNIX Systems Administrator, Linux Systems Design & Support (LSDS)
> Yale University Information Technology Services (ITS)
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080503/8dfeac58/attachment-0001.html 

From jsalbinson at arcastro.co.uk  Sun May  4 06:39:20 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Sun, 4 May 2008 14:39:20 +0100
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <481B0761.9050508@cognigencorp.com>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<200805020822.24352.jsalbinson@arcastro.co.uk>
	<481B0761.9050508@cognigencorp.com>
Message-ID: <200805041439.20793.jsalbinson@arcastro.co.uk>

Hi Darin,
I have traced the redhat-logos rpm to being a dependency on grub.
So my question is: Do we need grub to boot a diskless node?
Also do we need a password cracker and dictionary on a simple compute node?
Can we excise any more and shrink the initramdisk some more? I feel in a 
cutting mood....

Another question, where is the initial ramdisk set up? And what options are 
put in. My initramdisk/vnfs image won't run because of general elf settings.

Any thoughts? Thanks,
James Albinson

On Friday 02 May 2008, Darin Perusich wrote:
> I'm not sure but if you do 'rpm -qlif /path/to/image.png' it will tell
> you which package it's associated with.
>
> James Albinson wrote:
> > Hi Darin,
> > Thanks for the tip, but where is the logo stuff configured?
> > I had a scan through the vnfs setup file, and couldn't find the pointer
> > to the file that specifies the rpms  to use in the vnfs. Am I missing
> > something? (Probably yes...!).
> >
> > Chhers, and Thanks, JAmes Albinson
> >
> > On Thursday 01 May 2008, Darin Perusich wrote:
> >> You can safely remove them, all they're doing is bloating the image.
> >>
> >> James Albinson wrote:
> >>> Hi,
> >>> I notice that when building the SL 5.1 vfns image, some 30+Mb of Red
> >>> Hat Logos are downloaded.
> >>> Is this necessary? Out of 100Mb of download, its an awefully big chunk.
> >>> Maybe, if neeeded for copyright reasons, something really skinny could
> >>> be substituted?




From Darin.Perusich at cognigencorp.com  Mon May  5 05:13:20 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Mon, 05 May 2008 08:13:20 -0400
Subject: [Warewulf] Perceus vnfs SL 5.1 build
In-Reply-To: <200805041439.20793.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>	<200805020822.24352.jsalbinson@arcastro.co.uk>	<481B0761.9050508@cognigencorp.com>
	<200805041439.20793.jsalbinson@arcastro.co.uk>
Message-ID: <481EF9E0.3090308@cognigencorp.com>

Generally the VNFS images are pretty slim until you start adding stuff 
like compilers and those others tools need to make the machine useful. 
If you're really looking to shrink down the images take a look at 
"hybridizing" more of you're file systems, i.e. nfs mounting them. In 
/var/lib/perceus/vnfs/image the hybridize file controls which file 
systems are accessed over nfs. Mounting all of /usr via nfs will really 
cut down on the size of the image and give you back a lot of memory.

The /var/lib/perceus/image directory contains all the configuration 
params for your VNFS images.

James Albinson wrote:
> Hi Darin,
> I have traced the redhat-logos rpm to being a dependency on grub.
> So my question is: Do we need grub to boot a diskless node?
> Also do we need a password cracker and dictionary on a simple compute node?
> Can we excise any more and shrink the initramdisk some more? I feel in a 
> cutting mood....
> 
> Another question, where is the initial ramdisk set up? And what options are 
> put in. My initramdisk/vnfs image won't run because of general elf settings.
> 
> Any thoughts? Thanks,
> James Albinson
> 
> On Friday 02 May 2008, Darin Perusich wrote:
>> I'm not sure but if you do 'rpm -qlif /path/to/image.png' it will tell
>> you which package it's associated with.
>>
>> James Albinson wrote:
>>> Hi Darin,
>>> Thanks for the tip, but where is the logo stuff configured?
>>> I had a scan through the vnfs setup file, and couldn't find the pointer
>>> to the file that specifies the rpms  to use in the vnfs. Am I missing
>>> something? (Probably yes...!).
>>>
>>> Chhers, and Thanks, JAmes Albinson
>>>
>>> On Thursday 01 May 2008, Darin Perusich wrote:
>>>> You can safely remove them, all they're doing is bloating the image.
>>>>
>>>> James Albinson wrote:
>>>>> Hi,
>>>>> I notice that when building the SL 5.1 vfns image, some 30+Mb of Red
>>>>> Hat Logos are downloaded.
>>>>> Is this necessary? Out of 100Mb of download, its an awefully big chunk.
>>>>> Maybe, if neeeded for copyright reasons, something really skinny could
>>>>> be substituted?
> 
> 
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf

-- 
Darin Perusich
Unix Systems Administrator
Cognigen Corporation
395 Youngs Rd.
Williamsville, NY 14221
Phone: 716-633-3463
Email: darinper at cognigencorp.com


From Carsten.Bellon at bam.de  Mon May  5 08:13:10 2008
From: Carsten.Bellon at bam.de (Dr. Carsten Bellon)
Date: Mon, 05 May 2008 17:13:10 +0200
Subject: [Warewulf] getting Warewulf 3.0
Message-ID: <481F2406.7060600@bam.de>

I checked out Warewulf 3.0. But where is the ./configure file I have to run?

-- 
Dr.-Ing.         BAM Berlin * VIII.36 * D-12200 Berlin
Carsten Bellon   Tel/Fax: ++49 30 8104-3658 / -1837
                 Carsten.Bellon at bam.de



From jeff.blasius at yale.edu  Mon May  5 08:47:51 2008
From: jeff.blasius at yale.edu (Jeff Blasius)
Date: Mon, 5 May 2008 11:47:51 -0400
Subject: [Warewulf] Problems with tg3 in initramfs perceus 1.3.7 on RHEL
	5.1
In-Reply-To: <ae974e900805020943u7c8e058bvfec9b3569def2825@mail.gmail.com>
References: <bb83c5890805011404o467da50bnd69b1b0476fd2a0d@mail.gmail.com>
	<20080501214846.GO3941@kainx.org>
	<ae974e900805020943u7c8e058bvfec9b3569def2825@mail.gmail.com>
Message-ID: <bb83c5890805050847t38632da1s1724f26fef551c1a@mail.gmail.com>

Hello everybody. Thank You for your suggestions, but unfortunately
nothing seemed to work. I think at this point deadlines force me to
continue with the in-place solution.

I should add that if the owners had to pay the data center costs, the
cluster would have been gone three years ago (others in academia can
relate I'm sure). I have had plenty of success with redhat, tg3 and
perceus in the past.
Thanks,
                 jeff

For the record:
02:01.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5704
Gigabit Ethernet (rev 03)
02:01.1 Ethernet controller: Broadcom Corporation NetXtreme BCM5704
Gigabit Ethernet (rev 03)


On Fri, May 2, 2008 at 12:43 PM, Gary Leong <gwleong at gmail.com> wrote:
> I haven't seen this problem per se, but I can tell you if the machine
>  is a Dell unit with tg3, the Redhat kernel and its derivatives are
>  known to flip the interfaces, where eth0 is seen as the second
>  interface.  But I'm not sure if you're having problems with stage 1
>  with the perceus kernel or stage 2 with the second kernel.
>
>
>
>
>
>  On Thu, May 1, 2008 at 5:48 PM, Michael Jennings <mej at caoslinux.org> wrote:
>  > On Thursday, 01 May 2008, at 17:04:48 (-0400),
>  >
>  > Jeff Blasius wrote:
>  >
>  >  > I'm having some problems with the tg3 driver in the initramfs. The
>  >  > basic problem is that no link can be detected on the interface.
>  >  >
>  >  > The first ifconfig up does not show any signs of negotiating a link.
>  >  > If I cycle the interface down and up, using ifconfig in the init
>  >  > script, dmesg shows the link to be up at 1000Mb, but ethinfo -l still
>  >  > shows it to be down. If I set the ethinfo -l line in init to true,
>  >  > udhcpc still can not get an address... so there probably isn't a link
>  >  > even though dmesg says so.
>  >  >
>  >  > Has anyone come across this problem before especially with the tg3
>  >  > driver? Of course a RHEL install on the same hardware works fine.
>  >
>  >  While I can't say for sure what's breaking, I can say that we have
>  >  dozens of machines using the tg3 driver with Perceus 1.3.7 which are
>  >  functioning properly.  So Perceus is probably not the culprit here.
>  >
>  >  Michael
>  >
>  >  --
>  >  Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej at kainx.org>
>  >  Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
>  >  -----------------------------------------------------------------------
>  >   "A little learning is a dangerous thing;  Drink deep, or taste not
>  >   the Pierian spring:  There shallow draughts intoxicate the brain,
>  >   And drinking largely sobers us again."
>  >                             -- Alexander Pope, "An Essay on Criticism"
>  >
>  >
>  > _______________________________________________
>  >  Warewulf mailing list
>  >  Warewulf at caoslinux.org
>  >  http://lists.caosity.org/mailman/listinfo/warewulf
>  >
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Jeff Blasius / jeff.blasius at yale.edu
Phone: (203)432-9940 51 Prospect Rm. 011
High Performance Computing (HPC)
UNIX Systems Administrator, Linux Systems Design & Support (LSDS)
Yale University Information Technology Services (ITS)


From stevie_b_242 at yahoo.com  Mon May  5 11:19:08 2008
From: stevie_b_242 at yahoo.com (Steve Byrnes)
Date: Mon, 5 May 2008 11:19:08 -0700 (PDT)
Subject: [Warewulf] Ramdisks not supported with generic elf arguments
Message-ID: <328146.61893.qm@web51601.mail.re2.yahoo.com>

I ran into the issue where my nodes would not boot from a RHEL 5.1 VNFS and would fail with the message, "Ramdisks not supported with generic elf arguments."

This issue was solved for me by setting KEXEC_ARGS="--args-linux"
in the file /var/lib/perceus/vnfs/rhel-5.1-1.stateless.i386/config

There is even a comment in this config file that specifically says this is 
required when using a RHEL 5 kernel:

# KEXEC_ARGS: Any additional arguments that need to be set for the call to
# kexec (e.g. RHEL 5 uses a relocatable kernel thus it requires '--args-linux'
# to be used.

I would suggest that this issue and its resolution be called-out in the next edition of the Perceus User Guide.

People who are apparently much smarter than me have said;
(...I'm paraphrasing a few different sources here) 
-------------------------------------------------------------------------
In RHEL 5 and related kernels, an ELF header has been appended to relocatable bzImages and kexec therefore sees the boot-loader as an ELF file.  Relocatable kernels are required for use with the kdump / kexec facility.
Kdump is replacing diskdump and netdump for capturing vmcore files after a crash.
-------------------------------------------------------------------------

The above explanation is completely Greek to me but maybe it will help someone out there with a bigger brain than me.

(please feel free to correct or further explain all of this if you have more knowledge of this issue and its work-around)

BTW, my nodes still don't fully boot from my RHEL 5.1 VNFS but setting KEXEC_ARGS="--args-linux", at least got me past the "generic elf arguments" error.   %-)

--Steve



      ____________________________________________________________________________________
Be a better friend, newshound, and 
know-it-all with Yahoo! Mobile.  Try it now.  http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ


From jsalbinson at arcastro.co.uk  Wed May  7 06:06:11 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Wed, 7 May 2008 14:06:11 +0100
Subject: [Warewulf] Problems with vnfs configuration - thanks for replies
In-Reply-To: <mailman.1.1210075202.31462.warewulf@caoslinux.org>
References: <mailman.1.1210075202.31462.warewulf@caoslinux.org>
Message-ID: <200805071406.11701.jsalbinson@arcastro.co.uk>

Hello Darin,
I am deeply puzzled; /var/lib/perceus/image directory or file does not exist, 
nor the hybridize setup../
>
> Generally the VNFS images are pretty slim until you start adding stuff
> like compilers and those others tools need to make the machine useful.
> If you're really looking to shrink down the images take a look at
> "hybridizing" more of you're file systems, i.e. nfs mounting them. In
> /var/lib/perceus/vnfs/image the hybridize file controls which file
> systems are accessed over nfs. Mounting all of /usr via nfs will really
> cut down on the size of the image and give you back a lot of memory.
>
> The /var/lib/perceus/image directory contains all the configuration
> params for your VNFS images.

As for the kexec parameters, I am similarly stumped. I can't find these 
either?

Any thoughts?????????
Thanks, JAmes Albinson


From Darin.Perusich at cognigencorp.com  Wed May  7 06:13:13 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Wed, 07 May 2008 09:13:13 -0400
Subject: [Warewulf] Problems with vnfs configuration - thanks for replies
In-Reply-To: <200805071406.11701.jsalbinson@arcastro.co.uk>
References: <mailman.1.1210075202.31462.warewulf@caoslinux.org>
	<200805071406.11701.jsalbinson@arcastro.co.uk>
Message-ID: <4821AAE9.7030208@cognigencorp.com>

The 'image' in /var/lib/perceus/image is the name of the VNFS image, 
something like openSUSE-10.2.i386 or whatever you named it.

James Albinson wrote:
> Hello Darin,
> I am deeply puzzled; /var/lib/perceus/image directory or file does not exist, 
> nor the hybridize setup../
>> Generally the VNFS images are pretty slim until you start adding stuff
>> like compilers and those others tools need to make the machine useful.
>> If you're really looking to shrink down the images take a look at
>> "hybridizing" more of you're file systems, i.e. nfs mounting them. In
>> /var/lib/perceus/vnfs/image the hybridize file controls which file
>> systems are accessed over nfs. Mounting all of /usr via nfs will really
>> cut down on the size of the image and give you back a lot of memory.
>>
>> The /var/lib/perceus/image directory contains all the configuration
>> params for your VNFS images.
> 
> As for the kexec parameters, I am similarly stumped. I can't find these 
> either?
> 
> Any thoughts?????????
> Thanks, JAmes Albinson
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf

-- 
Darin Perusich
Unix Systems Administrator
Cognigen Corporation
395 Youngs Rd.
Williamsville, NY 14221
Phone: 716-633-3463
Email: darinper at cognigencorp.com


From gmkurtzer at gmail.com  Mon May 12 13:09:58 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Mon, 12 May 2008 13:09:58 -0700
Subject: [Warewulf] Presentation in Oakland tomorrow!
Message-ID: <571f1a060805121309q706e794av63a39c12c00b41ae@mail.gmail.com>

There will be a presentation in Oakland, California tomorrow (Tuesday
May 13th, 2008) for "Stateless Computing With Perceus" at the Open
Source Grid and Cluster Conference
(http://www.opensourcegridcluster.org/). Unfortunately it is not a
free conference, but if your in the area come by and say hello!

Thanks!
-- 
Greg Kurtzer
http://www.runlevelzero.net/
http://www.perceus.org/
http://www.caoslinux.org/


From laytonjb at charter.net  Mon May 12 13:16:17 2008
From: laytonjb at charter.net (Jeffrey B. Layton)
Date: Mon, 12 May 2008 16:16:17 -0400
Subject: [Warewulf] Presentation in Oakland tomorrow!
In-Reply-To: <571f1a060805121309q706e794av63a39c12c00b41ae@mail.gmail.com>
References: <571f1a060805121309q706e794av63a39c12c00b41ae@mail.gmail.com>
Message-ID: <4828A591.80107@charter.net>

Greg Kurtzer wrote:
> There will be a presentation in Oakland, California tomorrow (Tuesday
> May 13th, 2008) for "Stateless Computing With Perceus" at the Open
> Source Grid and Cluster Conference
> (http://www.opensourcegridcluster.org/). Unfortunately it is not a
> free conference, but if your in the area come by and say hello!
>
> Thanks!
>   
If you can - send me a copy!!! Look forward to seeing
you in a couple of weeks.

Thanks!

Jeff



From poknam at gmail.com  Thu May 29 01:31:47 2008
From: poknam at gmail.com (PN)
Date: Thu, 29 May 2008 16:31:47 +0800
Subject: [Warewulf] warewulf shows the wrong statistics
Message-ID: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>

Hi all,

i installed warewulf 3.0 r1781 in my compute nodes for monitoring.
however, it shows the wrong statistics:


[root at mgmt warewulf]# wwstats
--------------------------------------------------------------------------------
Total CPU utilization: 0%
          Total Nodes: 4
               Living: 4                           Warewulf
          Unavailable: 0                      Cluster Statistics
             Disabled: 0                 http://warewulf-cluster.org/
                Error: 0
                 Dead: 0
--------------------------------------------------------------------------------
 Node      Cluster        CPU       Memory (MB)      Swap (MB)      Current
 Name       Name       [util/num] [% used/total]   [% used/total]   Status
node0001                0%   (2)  % 82/2011        none
node0002                0%   (2)  % 47/2011        none
node0003                0%   (2)  % 48/2011        none
node0004                0%   (4)  % 83/2011        none

[root at node0001 ~]# free
             total       used       free     shared    buffers     cached
Mem:       2059556     580224    1479332          0        128     495884
-/+ buffers/cache:      84212    1975344
Swap:            0          0          0

I used 580 MB RAM, however it shows only 82MB are used.


[root at mgmt warewulf]# wwtop
Cluster totals: 4 nodes, 10 cpus, 25270 MHz, 7.86 GB mem
Avg:    0% cputil,  65.00 MB memutil, load 0.00,  84 procs, uptime   0
day(s)
High:   0% cputil,  83.00 MB memutil, load 0.00,  95 procs, uptime   0
day(s)
Low:    0% cputil,  47.00 MB memutil, load 0.00,  77 procs, uptime   0
day(s)
Node status:    4 ready,    0 unavailable,    0 down,    0 unknown
16:25:54 mgmt>
Node name    CPU  MEM SWAP Uptime   MHz    Arch Procs  Load  Net:KB/s
Stats/Util
node0001      0%   4%   0%   0.93  5320  x86_64   77   0.00         0 |
IDLE  |
node0002      0%   2%   0%   0.93  3990  x86_64   82   0.00         1 |
IDLE  |
node0003      0%   2%   0%   0.93  5320  x86_64   82   0.00         1 |
IDLE  |
node0004      0%   4%   0%   0.93 10640  x86_64   95   0.00         1 |
IDLE  |


1) % of memory used is wrong.
2) For node0001, i used Xeon 5130  @ 2.00GHz CPU, it shows 5320 MHz?
3) 77 Procs in node0001.....

Also, "wwsummary" also shows wrong % Mem Util for the localhost.


Regards,
PN
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080529/b47ea0cd/attachment.html 

From mej at caoslinux.org  Thu May 29 09:49:30 2008
From: mej at caoslinux.org (Michael Jennings)
Date: Thu, 29 May 2008 09:49:30 -0700
Subject: [Warewulf] warewulf shows the wrong statistics
In-Reply-To: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>
References: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>
Message-ID: <20080529164930.GA6471@kainx.org>

On Thursday, 29 May 2008, at 16:31:47 (+0800),
PN wrote:

> [root at mgmt warewulf]# wwstats
> --------------------------------------------------------------------------------
> Total CPU utilization: 0%
>           Total Nodes: 4
>                Living: 4                           Warewulf
>           Unavailable: 0                      Cluster Statistics
>              Disabled: 0                 http://warewulf-cluster.org/
>                 Error: 0
>                  Dead: 0
> --------------------------------------------------------------------------------
>  Node      Cluster        CPU       Memory (MB)      Swap (MB)      Current
>  Name       Name       [util/num] [% used/total]   [% used/total]   Status
> node0001                0%   (2)  % 82/2011        none
> node0002                0%   (2)  % 47/2011        none
> node0003                0%   (2)  % 48/2011        none
> node0004                0%   (4)  % 83/2011        none
> 
> [root at node0001 ~]# free
>              total       used       free     shared    buffers     cached
> Mem:       2059556     580224    1479332          0        128     495884
> -/+ buffers/cache:   >> 84212 << 1975344
> Swap:            0          0          0
> 
> I used 580 MB RAM, however it shows only 82MB are used.

That is correct.  Note the indicated figure above.  The "free" man
page will probably clarify this.

> [root at mgmt warewulf]# wwtop
> Cluster totals: 4 nodes, 10 cpus, 25270 MHz, 7.86 GB mem
> Avg:    0% cputil,  65.00 MB memutil, load 0.00,  84 procs, uptime   0
> day(s)
> High:   0% cputil,  83.00 MB memutil, load 0.00,  95 procs, uptime   0
> day(s)
> Low:    0% cputil,  47.00 MB memutil, load 0.00,  77 procs, uptime   0
> day(s)
> Node status:    4 ready,    0 unavailable,    0 down,    0 unknown
> 16:25:54 mgmt>
> Node name    CPU  MEM SWAP Uptime   MHz    Arch Procs  Load  Net:KB/s
> Stats/Util
> node0001      0%   4%   0%   0.93  5320  x86_64   77   0.00         0 |
> IDLE  |
> node0002      0%   2%   0%   0.93  3990  x86_64   82   0.00         1 |
> IDLE  |
> node0003      0%   2%   0%   0.93  5320  x86_64   82   0.00         1 |
> IDLE  |
> node0004      0%   4%   0%   0.93 10640  x86_64   95   0.00         1 |
> IDLE  |
> 
> 
> 1) % of memory used is wrong.

It is correct.

> 2) For node0001, i used Xeon 5130  @ 2.00GHz CPU, it shows 5320 MHz?

Dual core, right?

> 3) 77 Procs in node0001.....

Processes, not processors.

> Also, "wwsummary" also shows wrong % Mem Util for the localhost.

No, it is also correct.

Michael

-- 
Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej at kainx.org>
Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
-----------------------------------------------------------------------
 "I had rather men should ask why my statue is not set up, than why it
  is."        -- Cato the Elder (from Plutarch's "Political Precepts")


From poknam at gmail.com  Fri May 30 01:38:42 2008
From: poknam at gmail.com (PN)
Date: Fri, 30 May 2008 16:38:42 +0800
Subject: [Warewulf] warewulf shows the wrong statistics
In-Reply-To: <20080529164930.GA6471@kainx.org>
References: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>
	<20080529164930.GA6471@kainx.org>
Message-ID: <92daa7bf0805300138x2b015432x456e7f3faaf8d465@mail.gmail.com>

2008/5/30 Michael Jennings <mej at caoslinux.org>:

> On Thursday, 29 May 2008, at 16:31:47 (+0800),
> PN wrote:
>
> > [root at mgmt warewulf]# wwstats
> >
> --------------------------------------------------------------------------------
> > Total CPU utilization: 0%
> >           Total Nodes: 4
> >                Living: 4                           Warewulf
> >           Unavailable: 0                      Cluster Statistics
> >              Disabled: 0                 http://warewulf-cluster.org/
> >                 Error: 0
> >                  Dead: 0
> >
> --------------------------------------------------------------------------------
> >  Node      Cluster        CPU       Memory (MB)      Swap (MB)
>  Current
> >  Name       Name       [util/num] [% used/total]   [% used/total]
> Status
> > node0001                0%   (2)  % 82/2011        none
> > node0002                0%   (2)  % 47/2011        none
> > node0003                0%   (2)  % 48/2011        none
> > node0004                0%   (4)  % 83/2011        none
> >
> > [root at node0001 ~]# free
> >              total       used       free     shared    buffers     cached
> > Mem:       2059556     580224    1479332          0        128     495884
> > -/+ buffers/cache:   >> 84212 << 1975344
> > Swap:            0          0          0
> >
> > I used 580 MB RAM, however it shows only 82MB are used.
>
> That is correct.  Note the indicated figure above.  The "free" man
> page will probably clarify this.


Ah....I think the display is somewhat misleading. People see that 82/2011
may immediate think that there are at least 1900 MB can be used, but
actually not. Personally, I will concern the memory used up by the whole
system more.


>
>
> > [root at mgmt warewulf]# wwtop
> > Cluster totals: 4 nodes, 10 cpus, 25270 MHz, 7.86 GB mem
> > Avg:    0% cputil,  65.00 MB memutil, load 0.00,  84 procs, uptime   0
> > day(s)
> > High:   0% cputil,  83.00 MB memutil, load 0.00,  95 procs, uptime   0
> > day(s)
> > Low:    0% cputil,  47.00 MB memutil, load 0.00,  77 procs, uptime   0
> > day(s)
> > Node status:    4 ready,    0 unavailable,    0 down,    0 unknown
> > 16:25:54 mgmt>
> > Node name    CPU  MEM SWAP Uptime   MHz    Arch Procs  Load  Net:KB/s
> > Stats/Util
> > node0001      0%   4%   0%   0.93  5320  x86_64   77   0.00         0 |
> > IDLE  |
> > node0002      0%   2%   0%   0.93  3990  x86_64   82   0.00         1 |
> > IDLE  |
> > node0003      0%   2%   0%   0.93  5320  x86_64   82   0.00         1 |
> > IDLE  |
> > node0004      0%   4%   0%   0.93 10640  x86_64   95   0.00         1 |
> > IDLE  |
> >
> >
> > 1) % of memory used is wrong.
>
> It is correct.


As stated above, I would concern more about the memory used up by the whole
system.

>
>
> > 2) For node0001, i used Xeon 5130  @ 2.00GHz CPU, it shows 5320 MHz?
>
> Dual core, right?


Yes, 1 dual core cpu. How to come up the 5320 MHz?


>
>
> > 3) 77 Procs in node0001.....
>
> Processes, not processors.


That's fine.

Thanks for your reply.

Regards,
PN

>
>
> > Also, "wwsummary" also shows wrong % Mem Util for the localhost.
>
> No, it is also correct.
>
> Michael
>
> --
> Michael Jennings (a.k.a. KainX)  http://www.kainx.org/  <mej at kainx.org>
> Linux Server/Cluster Admin, LBL.gov       Author, Eterm (www.eterm.org)
> -----------------------------------------------------------------------
>  "I had rather men should ask why my statue is not set up, than why it
>  is."        -- Cato the Elder (from Plutarch's "Political Precepts")
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080530/774547d3/attachment.html 

From bernard at vanhpc.org  Fri May 30 12:17:58 2008
From: bernard at vanhpc.org (Bernard Li)
Date: Fri, 30 May 2008 12:17:58 -0700
Subject: [Warewulf] warewulf shows the wrong statistics
In-Reply-To: <92daa7bf0805300138x2b015432x456e7f3faaf8d465@mail.gmail.com>
References: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>
	<20080529164930.GA6471@kainx.org>
	<92daa7bf0805300138x2b015432x456e7f3faaf8d465@mail.gmail.com>
Message-ID: <d4c731da0805301217s3134f7c6v3901639be2ac6bd1@mail.gmail.com>

Hi PN:

On Fri, May 30, 2008 at 1:38 AM, PN <poknam at gmail.com> wrote:

>> > [root at node0001 ~]# free
>> >              total       used       free     shared    buffers
>> > cached
>> > Mem:       2059556     580224    1479332          0        128
>> > 495884
>> > -/+ buffers/cache:   >> 84212 << 1975344
>> > Swap:            0          0          0
>> >
>> > I used 580 MB RAM, however it shows only 82MB are used.
>>
>> That is correct.  Note the indicated figure above.  The "free" man
>> page will probably clarify this.
>
> Ah....I think the display is somewhat misleading. People see that 82/2011
> may immediate think that there are at least 1900 MB can be used, but
> actually not. Personally, I will concern the memory used up by the whole
> system more.

Warewulf considers cached and buffered memory free and available.
These are memory that are being used by disk caches and buffers that
could be freed up when they are needed.

In reality, I have seen stateless systems without swap invoke
OOM-killer even though "freeable" memory are available.  But perhaps
there are some tweaks to the virtual memory subsystem that should be
done on such a system to prevent this from happening.

Cheers,

Bernard


From stevie_b_242 at yahoo.com  Fri May 30 15:54:53 2008
From: stevie_b_242 at yahoo.com (Steve Byrnes)
Date: Fri, 30 May 2008 15:54:53 -0700 (PDT)
Subject: [Warewulf] Trouble creating/importing a new VNFS in Perceus
	1.4.0-1781
Message-ID: <558796.63225.qm@web51610.mail.re2.yahoo.com>

Greetings all.

I'm trying to get Perceus 1.4.0-1781 up-and-running on a small test cluster.  I previously had this setup running on 1.3.7 and I suspect that something is still left-over from the previous install and this is interfering with the operations of 1.4.0.

I previous just installed 1.4.0 over-the-top-of 1.3.7 but when I started to have issues creating and importing a vnfs in 1.4.0, I decided to uninstall the 1.3.7 RPM and run a make uninstall for 1.4.0, to try to get back to a clean base.

Can anyone explain why I am seeing the issue illustrated below and tell me how I might resolve it?
(I imagine I could just rename my vnfs so that it is not the same as the name I previously used under 1.3.7, but I wouldn't learn much or help make the software better by doing that %-)

Thanks!

--Steve

==========================================================================

[root at top vnfs-scripts]# ./chroot2stateless.sh /var/tmp/vnfs/rhel-5.1-1.i386 /root/rhel-5.1-1.stateless.i386.vnfs

Creating VNFS capsule 'rhel-5.1-1.stateless.i386.vnfs'
Setting default kernel to vmlinuz-2.6.18-53.el5
Kernel seems to be relocatable...
Building rootfs ...
Creating additional devices ...
Creating VNFS capsule scripts
Compressing capsule ...

WROTE: /root/rhel-5.1-1.stateless.i386.vnfs

[root at top lib]# perceus vnfs import /root/rhel-5.1-1.stateless.i386.vnfs
Importing 'rhel-5.1-1.stateless.i386'
ERROR: VNFS 'rhel-5.1-1.stateless.i386' is already installed!

[root at top lib]# perceus vnfs list
rhel-5.1-1.stateless.i386

[root at top lib]# perceus vnfs mount rhel-5.1-1.stateless.i386
VNFS 'rhel-5.1-1.stateless.i386' is already mounted!

[root at top lib]# perceus vnfs umount rhel-5.1-1.stateless.i386
ERROR: VNFS 'rhel-5.1-1.stateless.i386' is not mounted!

==========================================================================


      


From gmkurtzer at gmail.com  Fri May 30 16:06:03 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Fri, 30 May 2008 16:06:03 -0700
Subject: [Warewulf] Trouble creating/importing a new VNFS in Perceus
	1.4.0-1781
In-Reply-To: <558796.63225.qm@web51610.mail.re2.yahoo.com>
References: <558796.63225.qm@web51610.mail.re2.yahoo.com>
Message-ID: <571f1a060805301606n538eace8q6684992d5185bc6c@mail.gmail.com>

I would recommend to remove all traces of Perceus, rm -rf /etc/perceus
/var/lib/perceus; and any packages you have. This is not normally a
requirement for an update, but I think somehow you got yourself into a
situation with the VNFS version conflicts between 1.3 and 1.4 formats
(yes, they slightly changed) as well as the DB backends (although
there is a compatibility abstraction for the 1.3 DB).

Then start over making sure that you have recreated your VNFS capsule
using the new scripts, and try another import.

It is possible that due to the state of development of the 1.4 tree
there are some things not working quite right at this point and need
to be tracked down.

Lastly check out the --debug option and prepare for increased verbosity. ;)

Thanks,
Greg


On Fri, May 30, 2008 at 3:54 PM, Steve Byrnes <stevie_b_242 at yahoo.com> wrote:
> Greetings all.
>
> I'm trying to get Perceus 1.4.0-1781 up-and-running on a small test cluster.  I previously had this setup running on 1.3.7 and I suspect that something is still left-over from the previous install and this is interfering with the operations of 1.4.0.
>
> I previous just installed 1.4.0 over-the-top-of 1.3.7 but when I started to have issues creating and importing a vnfs in 1.4.0, I decided to uninstall the 1.3.7 RPM and run a make uninstall for 1.4.0, to try to get back to a clean base.
>
> Can anyone explain why I am seeing the issue illustrated below and tell me how I might resolve it?
> (I imagine I could just rename my vnfs so that it is not the same as the name I previously used under 1.3.7, but I wouldn't learn much or help make the software better by doing that %-)
>
> Thanks!
>
> --Steve
>
> ==========================================================================
>
> [root at top vnfs-scripts]# ./chroot2stateless.sh /var/tmp/vnfs/rhel-5.1-1.i386 /root/rhel-5.1-1.stateless.i386.vnfs
>
> Creating VNFS capsule 'rhel-5.1-1.stateless.i386.vnfs'
> Setting default kernel to vmlinuz-2.6.18-53.el5
> Kernel seems to be relocatable...
> Building rootfs ...
> Creating additional devices ...
> Creating VNFS capsule scripts
> Compressing capsule ...
>
> WROTE: /root/rhel-5.1-1.stateless.i386.vnfs
>
> [root at top lib]# perceus vnfs import /root/rhel-5.1-1.stateless.i386.vnfs
> Importing 'rhel-5.1-1.stateless.i386'
> ERROR: VNFS 'rhel-5.1-1.stateless.i386' is already installed!
>
> [root at top lib]# perceus vnfs list
> rhel-5.1-1.stateless.i386
>
> [root at top lib]# perceus vnfs mount rhel-5.1-1.stateless.i386
> VNFS 'rhel-5.1-1.stateless.i386' is already mounted!
>
> [root at top lib]# perceus vnfs umount rhel-5.1-1.stateless.i386
> ERROR: VNFS 'rhel-5.1-1.stateless.i386' is not mounted!
>
> ==========================================================================
>
>
>
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Greg Kurtzer
http://www.runlevelzero.net/
http://www.perceus.org/
http://www.caoslinux.org/


From poknam at gmail.com  Fri May 30 20:39:38 2008
From: poknam at gmail.com (PN)
Date: Sat, 31 May 2008 11:39:38 +0800
Subject: [Warewulf] warewulf shows the wrong statistics
In-Reply-To: <d4c731da0805301217s3134f7c6v3901639be2ac6bd1@mail.gmail.com>
References: <92daa7bf0805290131l55a91a72waef602085272d799@mail.gmail.com>
	<20080529164930.GA6471@kainx.org>
	<92daa7bf0805300138x2b015432x456e7f3faaf8d465@mail.gmail.com>
	<d4c731da0805301217s3134f7c6v3901639be2ac6bd1@mail.gmail.com>
Message-ID: <92daa7bf0805302039l4949bb5chb273196458db6cc2@mail.gmail.com>

Hi Bernard,

Yes, previously when I do some benchmark, some sofeware may suddenly used up
a lot of memory and cause OOM problem.
So do you mean that it is suggested to invoke a harddisk and make a swap on
it?
Regards,
PN


2008/5/31 Bernard Li <bernard at vanhpc.org>:

> Hi PN:
>
> On Fri, May 30, 2008 at 1:38 AM, PN <poknam at gmail.com> wrote:
>
> >> > [root at node0001 ~]# free
> >> >              total       used       free     shared    buffers
> >> > cached
> >> > Mem:       2059556     580224    1479332          0        128
> >> > 495884
> >> > -/+ buffers/cache:   >> 84212 << 1975344
> >> > Swap:            0          0          0
> >> >
> >> > I used 580 MB RAM, however it shows only 82MB are used.
> >>
> >> That is correct.  Note the indicated figure above.  The "free" man
> >> page will probably clarify this.
> >
> > Ah....I think the display is somewhat misleading. People see that 82/2011
> > may immediate think that there are at least 1900 MB can be used, but
> > actually not. Personally, I will concern the memory used up by the whole
> > system more.
>
> Warewulf considers cached and buffered memory free and available.
> These are memory that are being used by disk caches and buffers that
> could be freed up when they are needed.
>
> In reality, I have seen stateless systems without swap invoke
> OOM-killer even though "freeable" memory are available.  But perhaps
> there are some tweaks to the virtual memory subsystem that should be
> done on such a system to prevent this from happening.
>
> Cheers,
>
> Bernard
>  _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080531/fa4c590e/attachment.html