From poknam at gmail.com  Wed Apr  2 00:31:10 2008
From: poknam at gmail.com (PN)
Date: Wed, 2 Apr 2008 15:31:10 +0800
Subject: [Warewulf] 3 questions in perceus
In-Reply-To: <039501c890ee$98fd9170$cb00a8c0@terminal209>
References: <92daa7bf0803270300l3eb4db3dr9b61137e972f9e6b@mail.gmail.com>
	<006101c8902f$29ee39c0$cb00a8c0@terminal209>
	<92daa7bf0803280745m36defb42y2646a33c913d0d1c@mail.gmail.com>
	<039501c890ee$98fd9170$cb00a8c0@terminal209>
Message-ID: <92daa7bf0804020031h761bf032u8735d88b06ede9a9@mail.gmail.com>

2008/3/29, Arthur Stevens <astevens at gravitypark.com>:
>
>  1) cool. by default we run light.
> 2) Rapid Boot allows the full perceus client, ib drivers, etc to be
> embedded on the motherboard. Zepher is a different solution with lots of
> info available via google, and it supports the full perceus and can be used
> by a lot of vendors. Think of it as a compromise between embedded and a usb
> key. The Mellanox package I could not tell you about, you might give
> Mellanox a call. If I recall correctly, it lets you netboot over pxe but it
> is not the same as embedded perceus. The perceus embedded cards will have
> the full perceus client on them.
> 3) Looks like you have something else going on of which I do not have
> enough data to troubleshoot. Without even knowing the board vendor, it's
> hard to tell where to start. You might want to contact Infiscale if in a
> hurry with this one. Otherwise gather all your data and hit the list here
> with your pastebin ;)
>

Thanks for your reply.

My motherboard vendor is Iwill which use Broadcom Corporation NetXtreme
BCM5721 chipset. The driver used is tg3.
I found that the tg3.ko already comes with perceus kernel, under
/lib/modules/2.6.21-perceus/kernel/drivers/net. However, it seems that it
was not loaded properly.

I've tried to boot the compute node in debug mode. it shows:

Provisiong from 11.1.0.1...
Begining provisioning with debug level 1


Now provisioning: node0000
VNFS: GTC
Group: cluster
Node ID: 00:14:25:00:04:48

Mounting via NFS 11.1.0.1: /var/lib/perceus/
Downloading file via NFS: vnfs/GTC/vnfs.img
Downloading file via NFS: vnfs/GTC/rootfs//boot/vmlinuz-2.6.18-53.el5
Un-mounting 11.1.0.1:/var/lib/perceus
Un-loading device drivers:
 \-> ib_ipoib ata_piix ib_mthca tg3               <<<--- is it normal to
unload the tg3 driver here??
 Total provision time: 19s

Sleeping for 10 seconds for debug evaluation...
Ramdisks not supported with generic elf arguments
Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
 Requesting DHCP configuration via ib0 spoofing eth0: ERROR: Could not
determine hardware address of device 'eth0'
 Provisioning from 11.1.0.1...
Waiting 30 seconds, and rebooting...

Press [ENTER] to inturrupt reboot and get a shell.


After I get into the shell and run "ifconfig eth0", it shows:

ifconfig: eth0: error fetching interface information: Device not found

However, if I run the command "insmod tg3.ko" under
/lib/modules/2.6.21-perceus/kernel/drivers/net, the network driver comes up
and can communicate.

Also, i've tried to enable the perceus modprobe module and add a line to the
/etc/perceus/modules/modprobe

all: tg3 sd_mod xfs ext3 sunrpc nfs_acl lockd nfs

But the result is the same.

Any idea about it?
Thanks a lot.

PN


> hope this helps,
>
> Arthur
>
>
>
> .Sent on a Blackberry using BBpro Mail v2.
>
> ----- Original Message -----
> *From:* PN <poknam at gmail.com>
> *To:* Arthur Stevens <astevens at gravitypark.com> ; The Warewulf Cluster
> Toolkit <warewulf at caoslinux.org>
> *Sent:* Friday, March 28, 2008 7:45 AM
> *Subject:* Re: [Warewulf] 3 questions in perceus
>
>
> Hi,
>
> 2008/3/28, Arthur Stevens <astevens at gravitypark.com>:
>
> >  Replies to 1-3....
> >
> > 1) you can go into your confs and adjust that. we run skinny by default.
> >
>
> OK, I will try this later.
>
>  2) yes you are :) IB by default does not support net booting without our
> > love. You need our Rapid Boot Payload or our Zepher, or our embedded usb to
> > boot directly from IB. Also the new Perceus imbedded IB cards will support
> > that as well. 1 wire IB is a commercial offering at this time.
> >
>
> There is a Boot over IB software package provided by mellanox. Is Rapid
> Boot Payload or Zepher similar to that?
>
>
>
>
>
> >  3) You need to add your driver that is needed to the perceus
> > kernel/vnfs then restart stuff and all should be fine. Just add support for
> > your tg3. Infiscale offers that as a service if needed or you just don't
> > have the time.
> >
>
> It's ok for me to add the module. Actually the tg3 module comes along with
> the vnfs kernel. So do i simply edit the /etc/perceus/modules/modprobe is
> enough?
>
> It seems that perceus 1.3.6 and 1.3.7 are using the same perceus kernel.
> Is there any reason why the node cannot boot with the same vnfs image?
>
> Thanks a lot,
> PN
>
>
>
>
> >
> > Arthur
> >
> >  ----- Original Message -----
> > *From:* PN <poknam at gmail.com>
> > *To:* The Warewulf Cluster Toolkit <warewulf at caoslinux.org>
> > *Sent:* Thursday, March 27, 2008 3:00 AM
> > *Subject:* [Warewulf] 3 questions in perceus
> >
> >
> > hi all,
> >
> > i have 3 questions in perceus:
> >
> > 1) the syslog redirection seems not working, the default is redirected
> > to the master node, but i can't find any compute node information in the
> > master node's /var/log/messages, except the TFTP messages.
> >
> > 2) it seems that perceus can use IB as dhcp and provisioning, but after
> > i set the /etc/perceus/dnsmasq.conf using ib0 and restart perceus, the
> > compute node still cannot boot from IB. it just cannot find any dhcp server,
> > am i missing something?
> >
> > 3) when using perceus-1.3.6, the compute node can boot up sucessfully.
> > however in 1.3.7, the compute node cannot boot. it seems that the driver
> > is not working in the new version.
> > my compute node uses the tg3 network driver.
> >
> > Provisioning from 192.168.30.20     <= it is the internet IP, different
> > from version 1.3.6, which shows the internal IP 11.1.0.1.
> >
> >  Now provisiong: node0001
> > VNFS: GT8000
> > Group: cluster
> > Node ID: 00:14:25:00:04
> >
> > + cat /found_nics                          <= these messages did not
> > appear in perceus 1.3.6
> > + ifconfig eth0 down
> > + ifconfig eth1 down
> > + ifconfig ib0 down
> > + ifconfig ib1 down
> > + [ ! -f /sbin/detect ]
> > + . /etc/functions
> > + . /etc/initramfs.conf
> > + DEVS=eth0 eth1 eth2 eth3 eth4 eth5 eth6 eth7 ib0 ib1
> > + MAX_TRIES=5
> > + echo Un-loading device drivers:
> > + echo -ne \->
> >   \->+ unload_module scsi_mod
> > + grep -q ^scsi_mod /proc/modules
> > + PATH=/sbin rmmod scsi_mod
> > + unload_module ib_ipoib
> > + grep -q ^ib_ipoib /proc/modules
> > + PATH=/sbin rmmod ib_ipoib
> > + echo -n ib_ipoib
> >   ib_ipoib+ cat /etc/modulerc
> > + read i
> > + /sbin/detect -q
> > + read i
> > + unload_module uhci_hcd
> > + grep -q ^uhci_hcd /proc/modules
> > + read i
> > + unload_module uhci-hcd
> > + grep -q ^uhci-hcd /proc/modules
> > + read i
> >  + unload_module ehci-hcd
> > + grep -q ^ehci-hcd /proc/modules
> > + read i
> >  + unload_module ata_piix
> > + grep -q ^ata_piix /proc/modules
> > + PATH=/sbin rmmod ata_piix
> > + echo -n ata_piix
> >  ata_piix+ read i
> >  + unload_module piix
> > + grep -q ^piix /proc/modules
> > + PATH=/sbin remmod piix
> > + read i
> >  + unload_module ib_mthca
> > + grep -q ^ib_mthca /proc/modules
> > + PATH=/sbin rmmod ib_mthca
> > + echo -n ib_mthca
> >  ib_mthca+ read i
> > + unload_module tg3
> >  + grep -q ^tg3 /proc/modules
> > + PATH=/sbin rmmod tg3
> > + echo -n tg3
> >  tg3+ read i
> > + unload_module tg3
> >  + grep -q ^tg3 /proc/modules
> > + read i
> > + echo
> > ....
> >
> > thanks,
> > PN
> >
> > ------------------------------
> >
> > _______________________________________________
> > 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
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20080402/07e830ba/attachment.html 

From Darin.Perusich at cognigencorp.com  Thu Apr  3 07:27:48 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Thu, 03 Apr 2008 10:27:48 -0400
Subject: [Warewulf] HA Perceus servers
Message-ID: <47F4E964.4030206@cognigencorp.com>

Hello,

I'm wondering what others are doing to provide availability for Perceus 
servers in the case of a failure? Sections 2.2.3 in the user guide 
states is possible but doesn't get into any details other then using a 
remote state in section 2.4.2. My concern when nfs sharing the state 
information are the DB files in /var/lib/perceus/database/ and file 
locking, multiple servers need to write to the DB files. This also bring 
into questions the 'vnfs transfer master' value in perceus.conf, if you 
have multiple Perceus master servers what would this value be, the local 
Perceus host, all the Perceus server?

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


From astevens at gravitypark.com  Thu Apr  3 09:23:27 2008
From: astevens at gravitypark.com (Arthur Stevens)
Date: Thu, 3 Apr 2008 09:23:27 -0700
Subject: [Warewulf] HA Perceus servers
References: <47F4E964.4030206@cognigencorp.com>
Message-ID: <004001c895a7$0c7217e0$cb00a8c0@terminal209>

Abstractual, our commercial offering has more advanced HA abilities 
including failover, migration, and a slew of nifty tricks to make it so you 
can literally go unplug the control server and not skip a beat.

More documentation for doing this with Perceus will be available soon. We 
can also provide custom installs from Infiscale with extreme high 
availability in mind.

So yes it is done quite often, but currently nobody has the time to document 
how, hehe.

Arthur


----- Original Message ----- 
From: "Darin Perusich" <Darin.Perusich at cognigencorp.com>
To: "The Warewulf Cluster Toolkit" <warewulf at caoslinux.org>
Sent: Thursday, April 03, 2008 7:27 AM
Subject: [Warewulf] HA Perceus servers


> Hello,
>
> I'm wondering what others are doing to provide availability for Perceus
> servers in the case of a failure? Sections 2.2.3 in the user guide
> states is possible but doesn't get into any details other then using a
> remote state in section 2.4.2. My concern when nfs sharing the state
> information are the DB files in /var/lib/perceus/database/ and file
> locking, multiple servers need to write to the DB files. This also bring
> into questions the 'vnfs transfer master' value in perceus.conf, if you
> have multiple Perceus master servers what would this value be, the local
> Perceus host, all the Perceus server?
>
> -- 
> Darin Perusich
> Unix Systems Administrator
> Cognigen Corporation
> 395 Youngs Rd.
> Williamsville, NY 14221
> Phone: 716-633-3463
> Email: darinper at cognigencorp.com
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
> 




From Darin.Perusich at cognigencorp.com  Thu Apr  3 09:58:21 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Thu, 03 Apr 2008 12:58:21 -0400
Subject: [Warewulf] HA Perceus servers
In-Reply-To: <004001c895a7$0c7217e0$cb00a8c0@terminal209>
References: <47F4E964.4030206@cognigencorp.com>
	<004001c895a7$0c7217e0$cb00a8c0@terminal209>
Message-ID: <47F50CAD.1090500@cognigencorp.com>

Hi Arthur,

Arthur Stevens wrote:
> Abstractual, our commercial offering has more advanced HA abilities 
> including failover, migration, and a slew of nifty tricks to make it so you 
> can literally go unplug the control server and not skip a beat.

Are we talking and active-active or multi-master setup or 
active-passive? So I at least of some fail-over I'll likely go with an 
active-passive configuration using heartbeat and drdb which is was I'm 
currently doing. It may not be the most efficient way to utilize 
resource but it works well. Then once the documentation becomes 
available I can reevaluate and move to something else.

> More documentation for doing this with Perceus will be available soon. We 
> can also provide custom installs from Infiscale with extreme high 
> availability in mind.

Great, I'll be waiting for it!

> So yes it is done quite often, but currently nobody has the time to document 
> how, hehe.

Documentation always suffers when time is limited ;-).

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


From Darin.Perusich at cognigencorp.com  Mon Apr  7 06:36:23 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Mon, 07 Apr 2008 09:36:23 -0400
Subject: [Warewulf] perceus-1.3.7 build error
Message-ID: <47FA2357.8090407@cognigencorp.com>

I'm doing an rpmbuild of 1.3.7 and when building the detect module I get 
the following error which causing the compile to fail.

Making all in detect
make[3]: Entering directory 
`/usr/src/packages/BUILD/perceus-1.3.7/src/detect'
if gcc -DHAVE_CONFIG_H -I. -I. -I../../src    -I../../src/detect 
-I../../src/ -I../.. -O2 -g -m32 -march=i586 -mtune=i686 
-fmessage-length=0 -D_FORTIFY_SOURCE=2 -MT detect-detect.o -MD -MP -MF 
".deps/detect-detect.Tpo" -c -o detect-detect.o `test -f 'detect.c' || 
echo './'`detect.c; \
         then mv -f ".deps/detect-detect.Tpo" ".deps/detect-detect.Po"; 
else rm -f ".deps/detect-detect.Tpo"; exit 1; fi
detect.c:23:27: error: linux/pci_ids.h: No such file or directory
make[3]: *** [detect-detect.o] Error 1

pci_ids.h exists in 3rd_party/_work/kernel/linux-2.6.21.5/include/linux/ 
but it's not being included. Any thoughts on why this might be happening?

I'm building on OpenSUSE 10.3.

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


From nilssa at kth.se  Mon Apr  7 09:02:36 2008
From: nilssa at kth.se (nilssa at kth.se)
Date: Mon, 7 Apr 2008 18:02:36 +0200 (CEST)
Subject: [Warewulf] Warewulf: Nodegroup configuration
Message-ID: <52027.130.237.70.23.1207584156.squirrel@webmail.sys.kth.se>

I have a warewulf cluster (WW 2.6.2) consisting of 21 nodes. It is set up
in three nodegroups, with the intention that a typical user will have
access to only one nodegroup, according to definitions in
/etc/warewulf/nodes/nodegroup1/config, etc. If I run wwnodes --sync, all
nodes are synced. However, all nodes become accessible according to the
same rules, namely those of the nodegroup that was synced most recently.

If I run wwnodes --info node00nn, then the correct list of users is
displayed (according to the config-file), but I was expecting, for
instance, that wwmpirun would send jobs by user1 only to nodes in user1:s
nodegroup. Have I misunderstood something, or is something incorrectly set
up?

All nodes are running the same vnfs. Also, I run debian, and therefore
would prefer to stay with WW 2.6.2, now that (almost) everything is
working.

Thanks,

Nils




From timattox at open-mpi.org  Mon Apr  7 10:13:44 2008
From: timattox at open-mpi.org (Tim Mattox)
Date: Mon, 7 Apr 2008 13:13:44 -0400
Subject: [Warewulf] Warewulf: Nodegroup configuration
In-Reply-To: <52027.130.237.70.23.1207584156.squirrel@webmail.sys.kth.se>
References: <52027.130.237.70.23.1207584156.squirrel@webmail.sys.kth.se>
Message-ID: <ea86ce220804071013r5b56c52t396d214b15b86709@mail.gmail.com>

Hi Nils,
I hope someone else can chime in here, but since I have a few moments...
The nodegroup stuff in the last release Warewulf wasn't fully developed, as
far as I can remember.  You are likely just running into a bug. :-(
If you are able
to fix the perl-scripts, feel free to post a patch, and I would think Greg could
roll a new bugfix release, though most of his time I'm sure is now devoted
to Perceus, and the soon to be reborn Warewulf-as-cluster-monitoring-kit.

On Mon, Apr 7, 2008 at 12:02 PM,  <nilssa at kth.se> wrote:
> I have a warewulf cluster (WW 2.6.2) consisting of 21 nodes. It is set up
>  in three nodegroups, with the intention that a typical user will have
>  access to only one nodegroup, according to definitions in
>  /etc/warewulf/nodes/nodegroup1/config, etc. If I run wwnodes --sync, all
>  nodes are synced. However, all nodes become accessible according to the
>  same rules, namely those of the nodegroup that was synced most recently.
>
>  If I run wwnodes --info node00nn, then the correct list of users is
>  displayed (according to the config-file), but I was expecting, for
>  instance, that wwmpirun would send jobs by user1 only to nodes in user1:s
>  nodegroup. Have I misunderstood something, or is something incorrectly set
>  up?
>
>  All nodes are running the same vnfs. Also, I run debian, and therefore
>  would prefer to stay with WW 2.6.2, now that (almost) everything is
>  working.
>
>  Thanks,
>
>  Nils
>
>
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Tim Mattox, Ph.D. - http://homepage.mac.com/tmattox/
 tmattox at gmail.com || timattox at open-mpi.org
 I'm a bright... http://www.the-brights.net/


From nilssa at kth.se  Tue Apr  8 02:34:30 2008
From: nilssa at kth.se (nilssa at kth.se)
Date: Tue, 8 Apr 2008 11:34:30 +0200 (CEST)
Subject: [Warewulf] Warewulf: Nodegroup configuration
In-Reply-To: <ea86ce220804071013r5b56c52t396d214b15b86709@mail.gmail.com>
References: <52027.130.237.70.23.1207584156.squirrel@webmail.sys.kth.se>   
	<ea86ce220804071013r5b56c52t396d214b15b86709@mail.gmail.com>
Message-ID: <52364.130.237.70.23.1207647270.squirrel@webmail.sys.kth.se>

Hi Tim,

thanks for a quick response. I suspected that it may be a bug (or just not
developed yet). I have been digging into the perl-scripts and have a rough
idea of how things work right now. My problem was that it was unclear to
me how things were supposed to work. But your reply encourages me to
modify the appropriate script(s) and post my solution when the problem is
straightened out.

   - Nils

> Hi Nils,
> I hope someone else can chime in here, but since I have a few moments...
The nodegroup stuff in the last release Warewulf wasn't fully developed,
as
> far as I can remember.  You are likely just running into a bug. :-( If
you are able
> to fix the perl-scripts, feel free to post a patch, and I would think
Greg
> could
> roll a new bugfix release, though most of his time I'm sure is now
devoted
> to Perceus, and the soon to be reborn
Warewulf-as-cluster-monitoring-kit.
>
> On Mon, Apr 7, 2008 at 12:02 PM,  <nilssa at kth.se> wrote:
>> I have a warewulf cluster (WW 2.6.2) consisting of 21 nodes. It is set up
>>  in three nodegroups, with the intention that a typical user will have
access to only one nodegroup, according to definitions in
>>  /etc/warewulf/nodes/nodegroup1/config, etc. If I run wwnodes --sync,
>> all
>>  nodes are synced. However, all nodes become accessible according to
the
>>  same rules, namely those of the nodegroup that was synced most
>> recently.
>>  If I run wwnodes --info node00nn, then the correct list of users is
displayed (according to the config-file), but I was expecting, for
instance, that wwmpirun would send jobs by user1 only to nodes in
>> user1:s
>>  nodegroup. Have I misunderstood something, or is something incorrectly
>> set
>>  up?
>>  All nodes are running the same vnfs. Also, I run debian, and therefore
would prefer to stay with WW 2.6.2, now that (almost) everything is
working.
>>  Thanks,
>>  Nils
>>  _______________________________________________
>>  Warewulf mailing list
>>  Warewulf at caoslinux.org
>>  http://lists.caosity.org/mailman/listinfo/warewulf
>
>
>
> --
> Tim Mattox, Ph.D. - http://homepage.mac.com/tmattox/
>  tmattox at gmail.com || timattox at open-mpi.org
>  I'm a bright... http://www.the-brights.net/
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf
>






From Darin.Perusich at cognigencorp.com  Mon Apr 14 11:45:04 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Mon, 14 Apr 2008 14:45:04 -0400
Subject: [Warewulf] auto-provisioning and hostname matching
Message-ID: <4803A630.10109@cognigencorp.com>

Does or can perceus perform any type of MAC address to hostname matching 
when nodes are being automatically provisioned? It appears to me that 
perceus takes the MAC of an incoming host and assigns whatever the next 
available node### is without preforming any type of arp/hostname 
matching. I can see where this might be advantageous if the clients 
haven't been registered in any existing dns/dhcp infrastructure but that 
isn't the case.

I don't mind having to manually add new nodes in perceus, I was just 
hoping to eliminate a step in the process ;-).

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


From gwleong at gmail.com  Tue Apr 15 20:43:42 2008
From: gwleong at gmail.com (Gary Leong)
Date: Tue, 15 Apr 2008 23:43:42 -0400
Subject: [Warewulf] auto-provisioning and hostname matching
In-Reply-To: <4803A630.10109@cognigencorp.com>
References: <4803A630.10109@cognigencorp.com>
Message-ID: <ae974e900804152043p3bf75a33sfd5f89ac9025f39e@mail.gmail.com>

you can preassign the hostname according to mac address by doing
something like...

e.g.

perceus node add ?i 00:19:B9:E4:CB:0A blizzard


On Mon, Apr 14, 2008 at 2:45 PM, Darin Perusich
<Darin.Perusich at cognigencorp.com> wrote:
> Does or can perceus perform any type of MAC address to hostname matching
>  when nodes are being automatically provisioned? It appears to me that
>  perceus takes the MAC of an incoming host and assigns whatever the next
>  available node### is without preforming any type of arp/hostname
>  matching. I can see where this might be advantageous if the clients
>  haven't been registered in any existing dns/dhcp infrastructure but that
>  isn't the case.
>
>  I don't mind having to manually add new nodes in perceus, I was just
>  hoping to eliminate a step in the process ;-).
>
>  --
>  Darin Perusich
>  Unix Systems Administrator
>  Cognigen Corporation
>  395 Youngs Rd.
>  Williamsville, NY 14221
>  Phone: 716-633-3463
>  Email: darinper at cognigencorp.com
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>


From Darin.Perusich at cognigencorp.com  Wed Apr 16 04:49:14 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Wed, 16 Apr 2008 07:49:14 -0400
Subject: [Warewulf] auto-provisioning and hostname matching
In-Reply-To: <ae974e900804152043p3bf75a33sfd5f89ac9025f39e@mail.gmail.com>
References: <4803A630.10109@cognigencorp.com>
	<ae974e900804152043p3bf75a33sfd5f89ac9025f39e@mail.gmail.com>
Message-ID: <4805E7BA.6020208@cognigencorp.com>

This is what I'm currently doing, but the auto-provisioning mechanism 
does not, it adds the node as the next available unprovisioned node.

Gary Leong wrote:
> you can preassign the hostname according to mac address by doing
> something like...
> 
> e.g.
> 
> perceus node add ?i 00:19:B9:E4:CB:0A blizzard
> 
> 
> On Mon, Apr 14, 2008 at 2:45 PM, Darin Perusich
> <Darin.Perusich at cognigencorp.com> wrote:
>> Does or can perceus perform any type of MAC address to hostname matching
>>  when nodes are being automatically provisioned? It appears to me that
>>  perceus takes the MAC of an incoming host and assigns whatever the next
>>  available node### is without preforming any type of arp/hostname
>>  matching. I can see where this might be advantageous if the clients
>>  haven't been registered in any existing dns/dhcp infrastructure but that
>>  isn't the case.
>>
>>  I don't mind having to manually add new nodes in perceus, I was just
>>  hoping to eliminate a step in the process ;-).
>>
>>  --
>>  Darin Perusich
>>  Unix Systems Administrator
>>  Cognigen Corporation
>>  395 Youngs Rd.
>>  Williamsville, NY 14221
>>  Phone: 716-633-3463
>>  Email: darinper at cognigencorp.com
>>  _______________________________________________
>>  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

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



From travis.minor at wsm.com  Thu Apr 24 10:18:21 2008
From: travis.minor at wsm.com (Travis Minor)
Date: Thu, 24 Apr 2008 10:18:21 -0700
Subject: [Warewulf] Missing Cluster Address
Message-ID: <4810C0DD.9080603@wsm.com>

HI all,

I'm running into a weird problem where the Cluster network address is 
not being picked up by the system using WW 2.6.3 on CentOS 5.1.

I've defined the admin, sharedfs, and cluster networks in 
/etc/warewulf/nodes/default/"NODENAME", but after running wwnodes --sync 
my /etc/hosts file is missing the cluster network IP.
----------------------------------------------------------------------------------------------------------
# Node 'NODENAME' configuration (group: nodegroup1)
192.168.150.12	NODENAME-admin
	NODENAME NODENAME-cluster
192.168.150.12	NODENAME-sharedfs
----------------------------------------------------------------------------------------------------------


Similar results from wwnodes --info NODENAME
----------------------------------------------------------------------------------------------------------
wwnodes --info NODENAME

                        === NODE CONFIGIRATION ===

    Node name: NODENAME
  Description: Default node group 

   Node group: nodegroup1                       VNFS: default
       Kernel: kernel                       wwinitrd: wwinitrd.img
     Boot MAC: MACADDR              Secondary MAC:
        Users: gmk, god 

  User Groups: group1 


                       === NETWORK CONFIGURATION ===

    Admin dev: eth0          Admin address: 192.168.150.12/255.255.0.0
      SFS dev: eth0            SFS address: 192.168.150.12/255.255.0.0
  Cluster dev: eth0        Cluster address: /255.255.0.0


----------------------------------------------------------------------------------------------------------


The node configuration file specifies the cluster address as 
192.168.150.12 as well.
----------------------------------------------------------------------------------------------------------
cat /etc/warewulf/nodes/nodegroup1/NODENAME | grep cluster
     cluster  ipaddr   = 192.168.150.12
----------------------------------------------------------------------------------------------------------

I've replaced the /usr/lib/warewulf/modules/sync_defaults script with 
one from a working WW 2.6.3 cluster to no effect, I'm pretty sure I've 
just screwed up something simple I'm overlooking.

Any thoughts?

-Travis





From travis.minor at wsm.com  Thu Apr 24 10:34:28 2008
From: travis.minor at wsm.com (Travis Minor)
Date: Thu, 24 Apr 2008 10:34:28 -0700
Subject: [Warewulf] Missing Cluster Address
In-Reply-To: <4810C0DD.9080603@wsm.com>
References: <4810C0DD.9080603@wsm.com>
Message-ID: <4810C4A4.9040303@wsm.com>

Nevermind.

If any one's interested, the problem was here:

 > cat /etc/warewulf/nodes/nodegroup1/NODENAME | grep cluster
 >      cluster  ipaddr   = 192.168.150.12
                  ^
I had two spaces between cluster and address so the config was being 
ignored.

Thanks anyway.

Travis Minor wrote:
> HI all,
> 
> I'm running into a weird problem where the Cluster network address is 
> not being picked up by the system using WW 2.6.3 on CentOS 5.1.
> 
> I've defined the admin, sharedfs, and cluster networks in 
> /etc/warewulf/nodes/default/"NODENAME", but after running wwnodes --sync 
> my /etc/hosts file is missing the cluster network IP.
> ----------------------------------------------------------------------------------------------------------
> # Node 'NODENAME' configuration (group: nodegroup1)
> 192.168.150.12	NODENAME-admin
> 	NODENAME NODENAME-cluster
> 192.168.150.12	NODENAME-sharedfs
> ----------------------------------------------------------------------------------------------------------
> 
> 
> Similar results from wwnodes --info NODENAME
> ----------------------------------------------------------------------------------------------------------
> wwnodes --info NODENAME
> 
>                         === NODE CONFIGIRATION ===
> 
>     Node name: NODENAME
>   Description: Default node group 
> 
>    Node group: nodegroup1                       VNFS: default
>        Kernel: kernel                       wwinitrd: wwinitrd.img
>      Boot MAC: MACADDR              Secondary MAC:
>         Users: gmk, god 
> 
>   User Groups: group1 
> 
> 
>                        === NETWORK CONFIGURATION ===
> 
>     Admin dev: eth0          Admin address: 192.168.150.12/255.255.0.0
>       SFS dev: eth0            SFS address: 192.168.150.12/255.255.0.0
>   Cluster dev: eth0        Cluster address: /255.255.0.0
> 
> 
> ----------------------------------------------------------------------------------------------------------
> 
> 
> The node configuration file specifies the cluster address as 
> 192.168.150.12 as well.
> ----------------------------------------------------------------------------------------------------------
> cat /etc/warewulf/nodes/nodegroup1/NODENAME | grep cluster
>      cluster  ipaddr   = 192.168.150.12
> ----------------------------------------------------------------------------------------------------------
> 
> I've replaced the /usr/lib/warewulf/modules/sync_defaults script with 
> one from a working WW 2.6.3 cluster to no effect, I'm pretty sure I've 
> just screwed up something simple I'm overlooking.
> 
> Any thoughts?
> 
> -Travis
> 
> 
> 
> _______________________________________________
> Warewulf mailing list
> Warewulf at caoslinux.org
> http://lists.caosity.org/mailman/listinfo/warewulf

-- 
Travis Minor
Technical Engineer
travis.minor at wsm.com

Western Scientific, Inc
http://www.wsm.com
5444 Napa Street
San Diego, CA 92110
Tel 619.220.6580 x219
Fax 619.220.6590


From jsalbinson at arcastro.co.uk  Thu Apr 24 12:43:33 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Thu, 24 Apr 2008 20:43:33 +0100
Subject: [Warewulf] Suse 10.3 detect.c build failure - a solution... Also
	Warewulf 2.6.3 gzip build failure - a solution.
Message-ID: <200804242043.34418.jsalbinson@arcastro.co.uk>

Bug report for perceus 1.3.7 on Suse 10.3 i386 (not 64bit!).
make fails for detect.c as one of the required includes is missing 
from /usr/include/linux.
The following thump fixes it... (from ~werever/perceus.1.3.7)

cp ./3rd_party/_work/kernel/linux-2.6.21.5/include/linux/pci_ids.h /usr/include/linux

and than make performs as advertised...

My machine is a Piii 900MHz/768Mb
uname -a gives...
Linux cogito 2.6.22.17-0.1-default #1 SMP 2008/02/10 20:01:04 UTC i686 i686 
i386 GNU/Linux

Chhers, James Albinson

PS. A big thanks for putting up the warewulf 3 sources. These compile fine 
under Suse 10.3 (i386). I havn't tested an install yet...


From jsalbinson at arcastro.co.uk  Thu Apr 24 12:49:12 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Thu, 24 Apr 2008 20:49:12 +0100
Subject: [Warewulf] Problem with warewulf 2.6.3 - gzip.c has a typo ???
Message-ID: <200804242049.12820.jsalbinson@arcastro.co.uk>

Bug report for warewulf 2.6.3
Part of the make output...

gcc -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/include -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/include -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/libbb -Wall -Wstrict-prototypes -Wshadow -Os -march=i386 -mpreferred-stack-boundary=2 -falign-functions=0 -falign-jumps=0 -falign-loops=0 -fomit-frame-pointer -D_GNU_SOURCE -DNDEBUG     -c -o /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.o /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c
/home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:2146: warning: type 
qualifiers ignored on function return type
/home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:2146: error: 
conflicting types for 'build_bl_tree'
/home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:1623: error: 
previous declaration of 'build_bl_tree' was here
make[1]: *** [/home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.o] 
Error 1
make[1]: Leaving directory `/home/jsa/Astro/warewulf-2.6.3/src/busybox'
make: *** [wwinitrd] Error 2


Looking at gzip.c at the declaration of build_bl_tree, we see:

static void init_block(void);
static void pqdownheap(ct_data * tree, int k);
static void gen_bitlen(tree_desc * desc);
static void gen_codes(ct_data * tree, int max_code);
static int build_bl_tree(void);   <<<<<<<<<<<<<<<<<<<<<<<<<<HERE
static void scan_tree(ct_data * tree, int max_code);
static void send_tree(ct_data * tree, int max_code);
static int build_bl_tree(void);  <<<<<<<<<<<<<<<<<<<<<<<<< also here???Eh????? 
static void send_all_trees(int lcodes, int dcodes, int blcodes);
static void compress_block(ct_data * ltree, ct_data * dtree);
static void set_file_type(void);


Now further down see here...

/* ===========================================================================
 * Construct the Huffman tree for the bit lengths and return the index in
 * bl_order of the last bit length code to send.
 */
static const int build_bl_tree()   <<<<<<< 'const' has crept in +++++++++++++
{
        int max_blindex;        /* index of last bit length code of non zero 
freq */

******************************************************************
IF you zap the 'const' it all compiles...

Chhers, JAmes Albinson


From jsalbinson at arcastro.co.uk  Thu Apr 24 13:00:21 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Thu, 24 Apr 2008 21:00:21 +0100
Subject: [Warewulf] Warewulf gzip.c bug - an addendum...
Message-ID: <200804242100.21652.jsalbinson@arcastro.co.uk>

Run make in the src/...../archival directory  after you make the edit.

Otjherwise the gzip.c get refreshed from somewhere else and frustrates you.


Chhers, JAmes Albinson

too much coffee, too little sleep...


From gmkurtzer at gmail.com  Thu Apr 24 15:14:10 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Thu, 24 Apr 2008 15:14:10 -0700
Subject: [Warewulf] Suse 10.3 detect.c build failure - a solution...
	Also Warewulf 2.6.3 gzip build failure - a solution.
In-Reply-To: <200804242043.34418.jsalbinson@arcastro.co.uk>
References: <200804242043.34418.jsalbinson@arcastro.co.uk>
Message-ID: <571f1a060804241514h6a4a184dqe4d93c5b69ddd12d@mail.gmail.com>

Yes, this is a known issue with compiling on SuSE. For some reason
SuSE doesn't seem to include the pci_ids.h header in the glibc kernel
headers (/usr/include/linux/). Potential solutions:

- Crude hack and incorporate a pci_ids.h file from a recent kenrel in
the detect source (ewwwww)
- Document a workaround of copying a kernel source tree's pci_ids.h to
the right location
- Post a bug report with SuSE

If SuSE has another header that should be used in place of this, that
would be fine but as we don't develop on SuSE we would want someone to
follow up with that, or have SuSE check in with us (I heard some
rumors that they have some Perceus packages somewhere too).

Thanks for letting us know, and if you have any other ideas or
solutions, please let us know too.

Thanks!
Greg

On Thu, Apr 24, 2008 at 12:43 PM, James Albinson
<jsalbinson at arcastro.co.uk> wrote:
> Bug report for perceus 1.3.7 on Suse 10.3 i386 (not 64bit!).
>  make fails for detect.c as one of the required includes is missing
>  from /usr/include/linux.
>  The following thump fixes it... (from ~werever/perceus.1.3.7)
>
>  cp ./3rd_party/_work/kernel/linux-2.6.21.5/include/linux/pci_ids.h /usr/include/linux
>
>  and than make performs as advertised...
>
>  My machine is a Piii 900MHz/768Mb
>  uname -a gives...
>  Linux cogito 2.6.22.17-0.1-default #1 SMP 2008/02/10 20:01:04 UTC i686 i686
>  i386 GNU/Linux
>
>  Chhers, James Albinson
>
>  PS. A big thanks for putting up the warewulf 3 sources. These compile fine
>  under Suse 10.3 (i386). I havn't tested an install yet...
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Greg Kurtzer
http://www.runlevelzero.net/


From gmkurtzer at gmail.com  Thu Apr 24 15:20:06 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Thu, 24 Apr 2008 15:20:06 -0700
Subject: [Warewulf] Problem with warewulf 2.6.3 - gzip.c has a typo ???
In-Reply-To: <200804242049.12820.jsalbinson@arcastro.co.uk>
References: <200804242049.12820.jsalbinson@arcastro.co.uk>
Message-ID: <571f1a060804241520v69213dc2gfd6de4ffd310c4cb@mail.gmail.com>

All versions of Warewulf < 3 are no longer maintained, but this is a
good catch and may help some of our legacy users. :)

Thanks!
Greg


On Thu, Apr 24, 2008 at 12:49 PM, James Albinson
<jsalbinson at arcastro.co.uk> wrote:
> Bug report for warewulf 2.6.3
>  Part of the make output...
>
>  gcc -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/include -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/include -I/home/jsa/Astro/warewulf-2.6.3/src/busybox/libbb -Wall -Wstrict-prototypes -Wshadow -Os -march=i386 -mpreferred-stack-boundary=2 -falign-functions=0 -falign-jumps=0 -falign-loops=0 -fomit-frame-pointer -D_GNU_SOURCE -DNDEBUG     -c -o /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.o /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c
>  /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:2146: warning: type
>  qualifiers ignored on function return type
>  /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:2146: error:
>  conflicting types for 'build_bl_tree'
>  /home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.c:1623: error:
>  previous declaration of 'build_bl_tree' was here
>  make[1]: *** [/home/jsa/Astro/warewulf-2.6.3/src/busybox/archival/gzip.o]
>  Error 1
>  make[1]: Leaving directory `/home/jsa/Astro/warewulf-2.6.3/src/busybox'
>  make: *** [wwinitrd] Error 2
>
>
>  Looking at gzip.c at the declaration of build_bl_tree, we see:
>
>  static void init_block(void);
>  static void pqdownheap(ct_data * tree, int k);
>  static void gen_bitlen(tree_desc * desc);
>  static void gen_codes(ct_data * tree, int max_code);
>  static int build_bl_tree(void);   <<<<<<<<<<<<<<<<<<<<<<<<<<HERE
>  static void scan_tree(ct_data * tree, int max_code);
>  static void send_tree(ct_data * tree, int max_code);
>  static int build_bl_tree(void);  <<<<<<<<<<<<<<<<<<<<<<<<< also here???Eh?????
>  static void send_all_trees(int lcodes, int dcodes, int blcodes);
>  static void compress_block(ct_data * ltree, ct_data * dtree);
>  static void set_file_type(void);
>
>
>  Now further down see here...
>
>  /* ===========================================================================
>   * Construct the Huffman tree for the bit lengths and return the index in
>   * bl_order of the last bit length code to send.
>   */
>  static const int build_bl_tree()   <<<<<<< 'const' has crept in +++++++++++++
>  {
>         int max_blindex;        /* index of last bit length code of non zero
>  freq */
>
>  ******************************************************************
>  IF you zap the 'const' it all compiles...
>
>  Chhers, JAmes Albinson
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Greg Kurtzer
http://www.runlevelzero.net/


From gas at knc.ru  Fri Apr 25 05:55:20 2008
From: gas at knc.ru (Grigory Shamov)
Date: Fri, 25 Apr 2008 16:55:20 +0400
Subject: [Warewulf] Problem with building warewulf 2.6.3 on PowerPC
In-Reply-To: <200804242049.12820.jsalbinson@arcastro.co.uk>
References: <200804242049.12820.jsalbinson@arcastro.co.uk>
Message-ID: <4811D4B8.9060305@knc.ru>

Dear All,

I have trouble building warewuld 2.6 for PowerPC cluster (IBM pSeries). 
That is, busybox seem to have some Intel binaries there (conf and mconf 
things) which causes make to fail.

Did anybody built it for PowerPC? Any advice whether it is possible at 
all? Thank you!

I know that Warewulf is superceded by Perceus, but the latter seem to 
require nasm which is AFAIK not available for the Power arch.

--
WBR, Grigory Shamov
Kazan Science Centre of RAS
Kazan, Russia



From Darin.Perusich at cognigencorp.com  Fri Apr 25 05:57:21 2008
From: Darin.Perusich at cognigencorp.com (Darin Perusich)
Date: Fri, 25 Apr 2008 08:57:21 -0400
Subject: [Warewulf] Suse 10.3 detect.c build failure - a
 solution...	Also Warewulf 2.6.3 gzip build failure - a solution.
In-Reply-To: <571f1a060804241514h6a4a184dqe4d93c5b69ddd12d@mail.gmail.com>
References: <200804242043.34418.jsalbinson@arcastro.co.uk>
	<571f1a060804241514h6a4a184dqe4d93c5b69ddd12d@mail.gmail.com>
Message-ID: <4811D531.9090609@cognigencorp.com>

The work around for this is actually rather straight forward and only 
requires adding a single line to the perceus.spec file, let's not forget 
that Percues compiles the kernel during build ;-).

In the %configure section add the following line before %{__make} 
%{?mflags} and Perceus will build successfully. Granted if the kernel 
version changes the path will need to be updated but that's easily done. 
This is what I've done for my openSUSE 10.3 build.

%configure
ln -s ../../3rd_party/_work/kernel/linux-2.6.21.5/include/linux \
	src/detect
%{__make} %{?mflags}


I'll submit a bug report to SuSE's bugzilla about this header not being 
included. They've been getting a lot from me lately as I've been 
building my HA Perceus servers so what's one more!

Greg Kurtzer wrote:
> Yes, this is a known issue with compiling on SuSE. For some reason
> SuSE doesn't seem to include the pci_ids.h header in the glibc kernel
> headers (/usr/include/linux/). Potential solutions:
> 
> - Crude hack and incorporate a pci_ids.h file from a recent kenrel in
> the detect source (ewwwww)
> - Document a workaround of copying a kernel source tree's pci_ids.h to
> the right location
> - Post a bug report with SuSE
> 
> If SuSE has another header that should be used in place of this, that
> would be fine but as we don't develop on SuSE we would want someone to
> follow up with that, or have SuSE check in with us (I heard some
> rumors that they have some Perceus packages somewhere too).
> 
> Thanks for letting us know, and if you have any other ideas or
> solutions, please let us know too.
> 
> Thanks!
> Greg
> 
> On Thu, Apr 24, 2008 at 12:43 PM, James Albinson
> <jsalbinson at arcastro.co.uk> wrote:
>> Bug report for perceus 1.3.7 on Suse 10.3 i386 (not 64bit!).
>>  make fails for detect.c as one of the required includes is missing
>>  from /usr/include/linux.
>>  The following thump fixes it... (from ~werever/perceus.1.3.7)
>>
>>  cp ./3rd_party/_work/kernel/linux-2.6.21.5/include/linux/pci_ids.h /usr/include/linux
>>
>>  and than make performs as advertised...
>>
>>  My machine is a Piii 900MHz/768Mb
>>  uname -a gives...
>>  Linux cogito 2.6.22.17-0.1-default #1 SMP 2008/02/10 20:01:04 UTC i686 i686
>>  i386 GNU/Linux
>>
>>  Chhers, James Albinson
>>
>>  PS. A big thanks for putting up the warewulf 3 sources. These compile fine
>>  under Suse 10.3 (i386). I havn't tested an install yet...
>>  _______________________________________________
>>  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  Wed Apr 30 07:54:23 2008
From: Carsten.Bellon at bam.de (Dr. Carsten Bellon)
Date: Wed, 30 Apr 2008 16:54:23 +0200
Subject: [Warewulf] Newbie question
Message-ID: <4818881F.9040403@bam.de>

Im using Warewulf 2.6 quite a while . Now I try to use Perceus 1.3.7 for 
the first time (on my new TyanPSC T-600).

The head node is running CentOS5.1 and I have installed Perceus 1.3.7.

When starting the first compute node PXE on eth0 works fine. Then on the 
node perceus says:

Requesting DHCP configuration via ib0 spoofing eth0:<HWA of eth0>

ERROR: Could not obtain network configuration

Any help?
How I have to install/configure the infiniband network?
Can I force Perceus not to use infiniband for the moment?
Where can I check the dhcp configuration of Perceus?

Many thanks!
  Carsten

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



From Carsten.Bellon at bam.de  Wed Apr 30 09:21:51 2008
From: Carsten.Bellon at bam.de (Dr. Carsten Bellon)
Date: Wed, 30 Apr 2008 18:21:51 +0200
Subject: [Warewulf] Perceus 1.3.7: kernel/vnfs error
Message-ID: <48189C9F.1040905@bam.de>

I newly installed Perceus 1.3.7 on CentOS 5.1. While booting the first 
compute node I get:

Ramdisk not supported with generic elf arguments

Thanks for any help.
  Carsten

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



From Carsten.Bellon at bam.de  Wed Apr 30 10:50:10 2008
From: Carsten.Bellon at bam.de (Dr. Carsten Bellon)
Date: Wed, 30 Apr 2008 19:50:10 +0200
Subject: [Warewulf] Perceus 1.3.7: kernel/vnfs error
In-Reply-To: <48189C9F.1040905@bam.de>
References: <48189C9F.1040905@bam.de>
Message-ID: <4818B152.2060503@bam.de>

This problem is RHEL5 related. It needs KEXEC_ARGS="--args-linux" in
.../vnfs//config
Now provisoning works fine.
Sorry, for my silly question, which I have answered by myself herewith.
Thanks for Perceus!!!
  Carsten


Dr. Carsten Bellon wrote:
> I newly installed Perceus 1.3.7 on CentOS 5.1. While booting the first 
> compute node I get:
>
> Ramdisk not supported with generic elf arguments
>
> Thanks for any help.
>  Carsten
>

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



From gmkurtzer at gmail.com  Wed Apr 30 11:54:06 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 30 Apr 2008 11:54:06 -0700
Subject: [Warewulf] Newbie question
In-Reply-To: <4818881F.9040403@bam.de>
References: <4818881F.9040403@bam.de>
Message-ID: <571f1a060804301154y1b882805u8ba82066193ee583@mail.gmail.com>

It looks like it isn't finding your network device.

Get a debug shell, and type "ifconfig -a". Do any ethernet devices show up?

What kind of network card do you have in that node?


On Wed, Apr 30, 2008 at 7:54 AM, Dr. Carsten Bellon
<Carsten.Bellon at bam.de> wrote:
> Im using Warewulf 2.6 quite a while . Now I try to use Perceus 1.3.7 for
>  the first time (on my new TyanPSC T-600).
>
>  The head node is running CentOS5.1 and I have installed Perceus 1.3.7.
>
>  When starting the first compute node PXE on eth0 works fine. Then on the
>  node perceus says:
>
>  Requesting DHCP configuration via ib0 spoofing eth0:<HWA of eth0>
>
>  ERROR: Could not obtain network configuration
>
>  Any help?
>  How I have to install/configure the infiniband network?
>  Can I force Perceus not to use infiniband for the moment?
>  Where can I check the dhcp configuration of Perceus?
>
>  Many thanks!
>   Carsten
>
>  --
>  Dr.-Ing.         BAM Berlin * VIII.36 * D-12200 Berlin
>  Carsten Bellon   Tel/Fax: ++49 30 8104-3658 / -1837
>                  Carsten.Bellon at bam.de
>
>  _______________________________________________
>  Warewulf mailing list
>  Warewulf at caoslinux.org
>  http://lists.caosity.org/mailman/listinfo/warewulf
>



-- 
Greg Kurtzer
http://www.runlevelzero.net/


From jsalbinson at arcastro.co.uk  Wed Apr 30 12:52:28 2008
From: jsalbinson at arcastro.co.uk (James Albinson)
Date: Wed, 30 Apr 2008 20:52:28 +0100
Subject: [Warewulf] Perceus 1.3.7  - no response to node dhcp requests.
In-Reply-To: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
Message-ID: <200804302052.29161.jsalbinson@arcastro.co.uk>

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



From gmkurtzer at gmail.com  Wed Apr 30 17:05:49 2008
From: gmkurtzer at gmail.com (Greg Kurtzer)
Date: Wed, 30 Apr 2008 17:05:49 -0700
Subject: [Warewulf] Perceus 1.3.7 - no response to node dhcp requests.
In-Reply-To: <200804302052.29161.jsalbinson@arcastro.co.uk>
References: <mailman.1.1209211202.7606.warewulf@caoslinux.org>
	<200804302052.29161.jsalbinson@arcastro.co.uk>
Message-ID: <571f1a060804301705w1fbadd8diebb6b95fc1fcb9dd@mail.gmail.com>

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
>



-- 
Greg Kurtzer
http://www.runlevelzero.net/