From gmkurtzer at gmail.com  Sun Nov  9 12:30:10 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Sun, 9 Nov 2008 12:30:10 -0800
Subject: [Warewulf] Perceus 1.4.4 Released!
Message-ID: <571f1a060811091230h38ba5613q3dabc8d064e84525@mail.gmail.com>

The Infiscale team is happy to announce the release of Perceus 1.4.4.
This is a minor build and bugfix release in the 1.4 tree and can be
used to non-destructively update all the previous 1.4 releases. Major
changes include:

    * Some minor bug-fixes to the Perceus daemon
    * Use NFS as default while xget stabilizes (still avail, just not
default option)
    * Fixed linking error of libz into kexec-legacy
    * Set hostname in the xcpu module

Have fun,... but not too much fun. Its supposed to be work, remember?
-- 
Greg Kurtzer
http://www.infiscale.com/
http://www.runlevelzero.net/
http://www.perceus.org/
http://www.caoslinux.org/


From stefan at mdy.univie.ac.at  Wed Nov 12 06:18:59 2008
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 12 Nov 2008 15:18:59 +0100
Subject: [Warewulf] Question concerning network setup on nodes
Message-ID: <20081112141859.GA2731@loop.mdy.univie.ac.at>

I am probably making something incredibly stupid, but I am slightly at
a loss where to start looking. This is perceus 1.4.3 as set up by
caoslinux-nsa. The ipaddr module is activated (was activated by
default), and most everything should be set to default values as
provided throught the caos installer (sidekick).

If I understand the role of the ipaddr module correctly, it should
configure during boot a static network configuration on the fly.
Indeed, it gets the right info, but I end up (on the node) with
/etc/sysconfig/nics/eth0 *and* /etc/sysconfig/nics/eth1. The two are
identical, aside from the gateway line, which is empty for eth0. For
the time being, I prefer to be able to reach out from a node to the
outside world, but since /etc/sysconfig/eth0 contains an empty
GATEWAY= line, this doesn't work ...

Any ideas where I screwed up? More generally, where does the ipaddr
module (/var/lib/perceus//modules/ipaddr/nodescripts/50-ipaddr.pl) get
its default config info from ??

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  Wed Nov 12 06:36:45 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 12 Nov 2008 06:36:45 -0800
Subject: [Warewulf] Question concerning network setup on nodes
In-Reply-To: <20081112141859.GA2731@loop.mdy.univie.ac.at>
References: <20081112141859.GA2731@loop.mdy.univie.ac.at>
Message-ID: <571f1a060811120636qf5c82aanfd11eee16fb52862@mail.gmail.com>

It sounds like everything is working properly actually. Take a look at
/etc/perceus/modules/ipaddr.

The default option is to create a configuration for eth0 and eth1
(this is usually harmless) and catches issues with either eth0 *or*
eth1 only being connected to the LAN. The file *should* be self
documenting, but let me know if it isn't working as expected. :)

Thanks,
Greg


On Wed, Nov 12, 2008 at 6:18 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
> I am probably making something incredibly stupid, but I am slightly at
> a loss where to start looking. This is perceus 1.4.3 as set up by
> caoslinux-nsa. The ipaddr module is activated (was activated by
> default), and most everything should be set to default values as
> provided throught the caos installer (sidekick).
>
> If I understand the role of the ipaddr module correctly, it should
> configure during boot a static network configuration on the fly.
> Indeed, it gets the right info, but I end up (on the node) with
> /etc/sysconfig/nics/eth0 *and* /etc/sysconfig/nics/eth1. The two are
> identical, aside from the gateway line, which is empty for eth0. For
> the time being, I prefer to be able to reach out from a node to the
> outside world, but since /etc/sysconfig/eth0 contains an empty
> GATEWAY= line, this doesn't work ...
>
> Any ideas where I screwed up? More generally, where does the ipaddr
> module (/var/lib/perceus//modules/ipaddr/nodescripts/50-ipaddr.pl) get
> its default config info from ??
>
> 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
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



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


From stefan at mdy.univie.ac.at  Wed Nov 12 06:44:38 2008
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 12 Nov 2008 15:44:38 +0100
Subject: [Warewulf] Question concerning network setup on nodes
In-Reply-To: <571f1a060811120636qf5c82aanfd11eee16fb52862@mail.gmail.com>
References: <20081112141859.GA2731@loop.mdy.univie.ac.at>
	<571f1a060811120636qf5c82aanfd11eee16fb52862@mail.gmail.com>
Message-ID: <20081112144438.GG5928@loop.mdy.univie.ac.at>

On Wed, Nov 12, 2008 at 06:36:45AM -0800, Greg Kurtzer wrote:
> It sounds like everything is working properly actually. Take a look at
> /etc/perceus/modules/ipaddr.
> 
> The default option is to create a configuration for eth0 and eth1

OK ...

> (this is usually harmless) and catches issues with either eth0 *or*
> eth1 only being connected to the LAN. The file *should* be self
> documenting, but let me know if it isn't working as expected. :)

... so I understand why there is eth0 and eth1. Now, being poor, we
only have one ethernet (let alone any Infiniband) per node, so only
eth0 is read/used. However, since /etc/sysconfig/nics/eth0 contains

GATEWAY=

I don't get a default route out of the node LAN. Once I am in
production this may be OK (although we are all used to a "pool of more
or less fully configured workstations behind a gateway" rather than a
"tight cluster" with minimal systems on the nodes); at the moment,
it's definitely annoying.  As I said, /etc/sysconfig/nics/eth1 is
identical to eth0, with the exception of the DEVICE line, *and* it
contains GATEWAY={my_gateway_address} ...

Will try to understand ipaddr some more ...

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 Nov 12 06:47:32 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 12 Nov 2008 06:47:32 -0800
Subject: [Warewulf] Question concerning network setup on nodes
In-Reply-To: <20081112144438.GG5928@loop.mdy.univie.ac.at>
References: <20081112141859.GA2731@loop.mdy.univie.ac.at>
	<571f1a060811120636qf5c82aanfd11eee16fb52862@mail.gmail.com>
	<20081112144438.GG5928@loop.mdy.univie.ac.at>
Message-ID: <571f1a060811120647u37798243s34fd13f76d72f4d7@mail.gmail.com>

Lets say each node only needs eth0 and the default gateway is 192.168.1.1:

* eth0:[default]/[default]/192.168.1.1

Hope that helps!



On Wed, Nov 12, 2008 at 6:44 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
> On Wed, Nov 12, 2008 at 06:36:45AM -0800, Greg Kurtzer wrote:
>> It sounds like everything is working properly actually. Take a look at
>> /etc/perceus/modules/ipaddr.
>>
>> The default option is to create a configuration for eth0 and eth1
>
> OK ...
>
>> (this is usually harmless) and catches issues with either eth0 *or*
>> eth1 only being connected to the LAN. The file *should* be self
>> documenting, but let me know if it isn't working as expected. :)
>
> ... so I understand why there is eth0 and eth1. Now, being poor, we
> only have one ethernet (let alone any Infiniband) per node, so only
> eth0 is read/used. However, since /etc/sysconfig/nics/eth0 contains
>
> GATEWAY=
>
> I don't get a default route out of the node LAN. Once I am in
> production this may be OK (although we are all used to a "pool of more
> or less fully configured workstations behind a gateway" rather than a
> "tight cluster" with minimal systems on the nodes); at the moment,
> it's definitely annoying.  As I said, /etc/sysconfig/nics/eth1 is
> identical to eth0, with the exception of the DEVICE line, *and* it
> contains GATEWAY={my_gateway_address} ...
>
> Will try to understand ipaddr some more ...
>
> 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
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



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


From stefan at mdy.univie.ac.at  Wed Nov 12 08:35:00 2008
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 12 Nov 2008 17:35:00 +0100
Subject: [Warewulf] Question concerning network setup on nodes
In-Reply-To: <571f1a060811120647u37798243s34fd13f76d72f4d7@mail.gmail.com>
References: <20081112141859.GA2731@loop.mdy.univie.ac.at>
	<571f1a060811120636qf5c82aanfd11eee16fb52862@mail.gmail.com>
	<20081112144438.GG5928@loop.mdy.univie.ac.at>
	<571f1a060811120647u37798243s34fd13f76d72f4d7@mail.gmail.com>
Message-ID: <20081112163500.GH5928@loop.mdy.univie.ac.at>

On Wed, Nov 12, 2008 at 06:47:32AM -0800, Greg Kurtzer wrote:
> Lets say each node only needs eth0 and the default gateway is 192.168.1.1:
> 
> * eth0:[default]/[default]/192.168.1.1
> 
> Hope that helps!

Yes it did ... (I was busily staring at 50-ipaddr.pl without catching
that I should have looked at /etc/perceus/modules/ipaddr instead)

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 stefan at mdy.univie.ac.at  Tue Nov 18 04:03:40 2008
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Tue, 18 Nov 2008 13:03:40 +0100
Subject: [Warewulf] Mounting local filesystems on nodes
Message-ID: <20081118120340.GI9259@loop.mdy.univie.ac.at>


This is on perceus 1.4.4, clusterhead running 64bit caos-nsa-1.0rc?
(up to date), image provided to clients based on
caos-nsa-node-1.0rc1-1.stateless.x86_64

I want the nodes to have swap and a local "scratch" partition, hence I added 

/dev/sda3       none                    swap    defaults        0 0
/dev/sda4       /data                   ext3    defaults        1 2

to the fstab file in the vnfs. If my reading of the boot sequence is
right, then /etc/rc.sysinit should mount all local
filesystems (and turn on swap).  However, at the end of the boot,
neither /data nor swap. The boot error message tells me that the "special
device /dev/sda4 does not exist" ... On a completely booted machine
one can activate swap / mount /data without problems. 

Working around the problem is easy -- I could reissue the mount command
in rc.local (or somewhere at the end of the boot sequence) when presumably
devices have been created.  However, I am wondering whether this behavior
is intended, or whether this should be fixed differently. 

Thanks in advance,

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 Nov 18 09:23:09 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Tue, 18 Nov 2008 09:23:09 -0800
Subject: [Warewulf] Mounting local filesystems on nodes
In-Reply-To: <20081118120340.GI9259@loop.mdy.univie.ac.at>
References: <20081118120340.GI9259@loop.mdy.univie.ac.at>
Message-ID: <571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>

This maybe a silly question, but have the local devices been
partitioned and formated?

If so, do the devices show up post boot (meaning, can you log in and
do a "mount -a" and have everything work)?

Greg


On Tue, Nov 18, 2008 at 4:03 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
>
> This is on perceus 1.4.4, clusterhead running 64bit caos-nsa-1.0rc?
> (up to date), image provided to clients based on
> caos-nsa-node-1.0rc1-1.stateless.x86_64
>
> I want the nodes to have swap and a local "scratch" partition, hence I added
>
> /dev/sda3       none                    swap    defaults        0 0
> /dev/sda4       /data                   ext3    defaults        1 2
>
> to the fstab file in the vnfs. If my reading of the boot sequence is
> right, then /etc/rc.sysinit should mount all local
> filesystems (and turn on swap).  However, at the end of the boot,
> neither /data nor swap. The boot error message tells me that the "special
> device /dev/sda4 does not exist" ... On a completely booted machine
> one can activate swap / mount /data without problems.
>
> Working around the problem is easy -- I could reissue the mount command
> in rc.local (or somewhere at the end of the boot sequence) when presumably
> devices have been created.  However, I am wondering whether this behavior
> is intended, or whether this should be fixed differently.
>
> Thanks in advance,
>
> 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
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



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


From stefan at mdy.univie.ac.at  Wed Nov 19 00:07:22 2008
From: stefan at mdy.univie.ac.at (stefan at mdy.univie.ac.at)
Date: Wed, 19 Nov 2008 09:07:22 +0100 (CET)
Subject: [Warewulf] Mounting local filesystems on nodes
In-Reply-To: <571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>
References: <20081118120340.GI9259@loop.mdy.univie.ac.at>
	<571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>
Message-ID: <36641.62.47.152.118.1227082042.squirrel@www.mdy.univie.ac.at>

> This maybe a silly question, but have the local devices been
> partitioned and formated?

Yes

> If so, do the devices show up post boot (meaning, can you log in and
> do a "mount -a" and have everything work)?

and yes. I am reasonably sure that I could do a mount -a and swapon -a
at the end of the boot process and it would work. I haven't tried because
it "doesn't feel right" ...

Thanks,

Stefan

PS: The only nonstandard thing I can think of that I installed the
2.6.27.4 kernel for which you sent me recently the src.rpm



From stefan at mdy.univie.ac.at  Wed Nov 19 01:16:48 2008
From: stefan at mdy.univie.ac.at (Stefan Boresch)
Date: Wed, 19 Nov 2008 10:16:48 +0100
Subject: [Warewulf] Mounting local filesystems on nodes
In-Reply-To: <571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>
References: <20081118120340.GI9259@loop.mdy.univie.ac.at>
	<571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>
Message-ID: <20081119091648.GK9259@loop.mdy.univie.ac.at>

Greg,

I think I see where the problem probably lies.  rc.sysinit only
waits for udev when 

	if [ ! -f "/fastboot" -a ! -f "/.stateless" ]; then
		echo -n "   Waiting for block devices to come online..."
		etc...

and this test certainly evaluates to false for me ...

Since I *am* stateless and I see no problem otherwise with fastboot,

I went ahead and put the mount -a and swapon -a into 
rc.local (I certainly don't need swap and /data earlier, and since it's
scratch space I don't care if things get corrupted too much ...), and,
as expected, this works. 

Best regards,

Stefan

On Tue, Nov 18, 2008 at 09:23:09AM -0800, Greg Kurtzer wrote:
> This maybe a silly question, but have the local devices been
> partitioned and formated?
> 
> If so, do the devices show up post boot (meaning, can you log in and
> do a "mount -a" and have everything work)?
> 
> Greg
> 
> 
> On Tue, Nov 18, 2008 at 4:03 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
> >
> > This is on perceus 1.4.4, clusterhead running 64bit caos-nsa-1.0rc?
> > (up to date), image provided to clients based on
> > caos-nsa-node-1.0rc1-1.stateless.x86_64
> >
> > I want the nodes to have swap and a local "scratch" partition, hence I added
> >
> > /dev/sda3       none                    swap    defaults        0 0
> > /dev/sda4       /data                   ext3    defaults        1 2
> >
> > to the fstab file in the vnfs. If my reading of the boot sequence is
> > right, then /etc/rc.sysinit should mount all local
> > filesystems (and turn on swap).  However, at the end of the boot,
> > neither /data nor swap. The boot error message tells me that the "special
> > device /dev/sda4 does not exist" ... On a completely booted machine
> > one can activate swap / mount /data without problems.
> >
> > Working around the problem is easy -- I could reissue the mount command
> > in rc.local (or somewhere at the end of the boot sequence) when presumably
> > devices have been created.  However, I am wondering whether this behavior
> > is intended, or whether this should be fixed differently.
> >
> > Thanks in advance,
> >
> > 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
> > _______________________________________________
> > Warewulf mailing list
> > Warewulf at caoslinux.org
> > http://lists.caosity.org/mailman/listinfo/warewulf
> >
> 
> 
> 
> -- 
> Greg Kurtzer
> http://www.infiscale.com/
> http://www.runlevelzero.net/
> http://www.perceus.org/
> http://www.caoslinux.org/
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
> 

-- 
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 Nov 19 06:49:03 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 19 Nov 2008 06:49:03 -0800
Subject: [Warewulf] Mounting local filesystems on nodes
In-Reply-To: <20081119091648.GK9259@loop.mdy.univie.ac.at>
References: <20081118120340.GI9259@loop.mdy.univie.ac.at>
	<571f1a060811180923i45d81e1bk9c97f12cb7ba3685@mail.gmail.com>
	<20081119091648.GK9259@loop.mdy.univie.ac.at>
Message-ID: <571f1a060811190649x67f491e4kc2c4683ff363bf1b@mail.gmail.com>

Awesome. I will tweak up the rc.sysinit so that it doesn't exclude non
root file systems when running stateless. Good catch!

Greg

On Wed, Nov 19, 2008 at 1:16 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
> Greg,
>
> I think I see where the problem probably lies.  rc.sysinit only
> waits for udev when
>
>        if [ ! -f "/fastboot" -a ! -f "/.stateless" ]; then
>                echo -n "   Waiting for block devices to come online..."
>                etc...
>
> and this test certainly evaluates to false for me ...
>
> Since I *am* stateless and I see no problem otherwise with fastboot,
>
> I went ahead and put the mount -a and swapon -a into
> rc.local (I certainly don't need swap and /data earlier, and since it's
> scratch space I don't care if things get corrupted too much ...), and,
> as expected, this works.
>
> Best regards,
>
> Stefan
>
> On Tue, Nov 18, 2008 at 09:23:09AM -0800, Greg Kurtzer wrote:
>> This maybe a silly question, but have the local devices been
>> partitioned and formated?
>>
>> If so, do the devices show up post boot (meaning, can you log in and
>> do a "mount -a" and have everything work)?
>>
>> Greg
>>
>>
>> On Tue, Nov 18, 2008 at 4:03 AM, Stefan Boresch <stefan at mdy.univie.ac.at> wrote:
>> >
>> > This is on perceus 1.4.4, clusterhead running 64bit caos-nsa-1.0rc?
>> > (up to date), image provided to clients based on
>> > caos-nsa-node-1.0rc1-1.stateless.x86_64
>> >
>> > I want the nodes to have swap and a local "scratch" partition, hence I added
>> >
>> > /dev/sda3       none                    swap    defaults        0 0
>> > /dev/sda4       /data                   ext3    defaults        1 2
>> >
>> > to the fstab file in the vnfs. If my reading of the boot sequence is
>> > right, then /etc/rc.sysinit should mount all local
>> > filesystems (and turn on swap).  However, at the end of the boot,
>> > neither /data nor swap. The boot error message tells me that the "special
>> > device /dev/sda4 does not exist" ... On a completely booted machine
>> > one can activate swap / mount /data without problems.
>> >
>> > Working around the problem is easy -- I could reissue the mount command
>> > in rc.local (or somewhere at the end of the boot sequence) when presumably
>> > devices have been created.  However, I am wondering whether this behavior
>> > is intended, or whether this should be fixed differently.
>> >
>> > Thanks in advance,
>> >
>> > 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
>> > _______________________________________________
>> > Warewulf mailing list
>> > Warewulf at caoslinux.org
>> > http://lists.caosity.org/mailman/listinfo/warewulf
>> >
>>
>>
>>
>> --
>> Greg Kurtzer
>> http://www.infiscale.com/
>> http://www.runlevelzero.net/
>> http://www.perceus.org/
>> http://www.caoslinux.org/
>> _______________________________________________
>> Warewulf mailing list
>> Warewulf at caoslinux.org
>> http://lists.caosity.org/mailman/listinfo/warewulf
>>
>
> --
> Stefan Boresch
> Institute for Computational Biological Chemistry
> University of Vienna, Waehringerstr. 17       A-1090 Vienna, Austria
> Phone: -43-1-427752715                        Fax:   -43-1-427752790
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



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


From pgreer at structbio.pitt.edu  Wed Nov 19 11:18:20 2008
From: pgreer at structbio.pitt.edu (Phil Greer)
Date: Wed, 19 Nov 2008 14:18:20 -0500
Subject: [Warewulf] user account problems
Message-ID: <961DEC91-A9D5-45C2-B116-D785E780F9BC@structbio.pitt.edu>

OK, really weird problem here. I made several accounts a while back on  
my cluster,
but only recently setup the ssh auth keys for them.  they all work  
fine and I can log into
the nodes with no password. all new accounts that I have made lately  
are not working
with the ssh authkeys. even though everything has been done  
identically. or so I think.
is there some step tat has to be done after local accounts are created  
on the head node
before they will be able to log into the compute nodes?

btw. perceus 1.4.1, masterauth module on

if I log into the node as root I can see the updated passwd file with  
the new accounts,
and su to them. but there is no shadow file, so typeing the password  
is useless, and
paswordless login doesn't work what have I done wrong?

-Phil Greer



From pgreer at structbio.pitt.edu  Wed Nov 19 11:42:21 2008
From: pgreer at structbio.pitt.edu (Phil Greer)
Date: Wed, 19 Nov 2008 14:42:21 -0500
Subject: [Warewulf] user account problems
In-Reply-To: <961DEC91-A9D5-45C2-B116-D785E780F9BC@structbio.pitt.edu>
References: <961DEC91-A9D5-45C2-B116-D785E780F9BC@structbio.pitt.edu>
Message-ID: <02AA1660-6ED2-44F1-9D63-8CE9BA8D5E87@structbio.pitt.edu>

nevermind I was using the commands at the back of the perceus user  
guide.

you all need to ass to chmod 700 the .ssh directory...

-Phil Greer

On Nov 19, 2008, at 2:18 PM, Phil Greer wrote:

> OK, really weird problem here. I made several accounts a while back on
> my cluster,
> but only recently setup the ssh auth keys for them.  they all work
> fine and I can log into
> the nodes with no password. all new accounts that I have made lately
> are not working
> with the ssh authkeys. even though everything has been done
> identically. or so I think.
> is there some step tat has to be done after local accounts are created
> on the head node
> before they will be able to log into the compute nodes?
>
> btw. perceus 1.4.1, masterauth module on
>
> if I log into the node as root I can see the updated passwd file with
> the new accounts,
> and su to them. but there is no shadow file, so typeing the password
> is useless, and
> paswordless login doesn't work what have I done wrong?
>
> -Phil Greer
>
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



From pgreer at structbio.pitt.edu  Wed Nov 19 11:42:21 2008
From: pgreer at structbio.pitt.edu (Phil Greer)
Date: Wed, 19 Nov 2008 14:42:21 -0500
Subject: [Warewulf] user account problems
In-Reply-To: <961DEC91-A9D5-45C2-B116-D785E780F9BC@structbio.pitt.edu>
References: <961DEC91-A9D5-45C2-B116-D785E780F9BC@structbio.pitt.edu>
Message-ID: <02AA1660-6ED2-44F1-9D63-8CE9BA8D5E87@structbio.pitt.edu>

nevermind I was using the commands at the back of the perceus user  
guide.

you all need to ass to chmod 700 the .ssh directory...

-Phil Greer

On Nov 19, 2008, at 2:18 PM, Phil Greer wrote:

> OK, really weird problem here. I made several accounts a while back on
> my cluster,
> but only recently setup the ssh auth keys for them.  they all work
> fine and I can log into
> the nodes with no password. all new accounts that I have made lately
> are not working
> with the ssh authkeys. even though everything has been done
> identically. or so I think.
> is there some step tat has to be done after local accounts are created
> on the head node
> before they will be able to log into the compute nodes?
>
> btw. perceus 1.4.1, masterauth module on
>
> if I log into the node as root I can see the updated passwd file with
> the new accounts,
> and su to them. but there is no shadow file, so typeing the password
> is useless, and
> paswordless login doesn't work what have I done wrong?
>
> -Phil Greer
>
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>



From gmkurtzer at gmail.com  Tue Nov 25 11:03:49 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Tue, 25 Nov 2008 11:03:49 -0800
Subject: [Warewulf] Caos NSA 1.0 Released!
Message-ID: <571f1a060811251103r1489eb6bt29827d7bef1ab6f0@mail.gmail.com>

I realize this isn't the Caos announcements list, but because Caos NSA
1.0 comes with a fully integrated Perceus and Warewulf stack that is
Intel Cluster Ready certified, really focused on HPC, and co-developed
by many of the Perceus developers, I figured I can get away with a
quick plug. :)

http://lists.caosity.org/pipermail/caos/2008-November/002537.html
http://www.caoslinux.org/

Thanks!

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