Thanks for the update.
On 4/26/07, Greenseid, Joseph M. <[EMAIL PROTECTED]> wrote:
For the sake of closure, I figured I'd come back to this to update on what
I found.
My OS is CentOS 4.4. I think that it is actually a network issue, not an
NFS issue. I found some similar problems on the CentOS mailing list
referring to this.
Anyway, in my testing, I threw a 10 packet ping test into the netfs init
script, and though all packets get dropped, it appears to be enough to let
the network sort itself out, so NFS is now mounting the remote file
systems. It's very bizarre, and I will continue investigating, but at least
it's I've figured out where the problem is and it's mostly working now...
--Joe
________________________________
From: [EMAIL PROTECTED] on behalf of Michael
Edwards
Sent: Mon 4/16/2007 9:44 AM
To: oscar-users@lists.sourceforge.net
Subject: Re: [Oscar-users] NFS file systems not mounting on boot
Have you tried turning off pfilter? I hadn't noticed you were running
4.2.1, but the pfilter package has problems on some systems. I don't
think they were with NFS, but since this appears to be some sort of boot
timing issue you might try making a new image which doesn't include it and
push that to the nodes. You'll need to turn the service off on the head
node as well.
You might try a more specific NFS list, OSCAR doesn't do much with the
stock NFS install to my knowledge so the fact that they are unfamiliar with
OSCAR shouldn't mater much.
How many nodes are you using? Have you tested with only one node to see
if it is a network bottleneck of some kind? My brain keeps straying toward
the switch, or service startup order, since everything works once everything
is going.
On 4/16/07, Greenseid, Joseph M. <[EMAIL PROTECTED]> wrote:
I made this change but it does not seem to help. The nodes still
don't catch the NFS file systems on boot. Very strange...
--Joe
________________________________
From: [EMAIL PROTECTED] on behalf of
Greenseid, Joseph M.
Sent: Fri 4/13/2007 8:43 AM
To: oscar-users@lists.sourceforge.net;
oscar-users@lists.sourceforge.net
Subject: Re: [Oscar-users] NFS file systems not mounting on boot
Thanks for the suggestion. I'll give it a try and let you know how
it goes.
--Joe
________________________________
From: [EMAIL PROTECTED] on behalf of
Michael Edwards
Sent: Thu 4/12/2007 5:54 PM
To: oscar-users@lists.sourceforge.net
Subject: Re: [Oscar-users] NFS file systems not mounting on boot
One thing you can try is to kick the value of the System V startup
script toward the end of the boot process. I have had occasional problems
where the network startup took a very long time but seemed to go on in the
background after the system said [ok] and there was not any connectivity
until much later in the boot process.
check in /var/lib/systemimager/images/<imagename>/etc/rc3.d/ for
netfs and nfslock (at least), they should show up as symlinks, something
like S25netfs and S14nfslock. You can mv the file to something like
S90netfs and S90nfslock. You may need to play with this, but if nfslock is
broken things should work, just with lots of errors in the log files.
Don't forget to either cpush the files to the /etc/rc3.d on the
nodes and/or reimage the nodes.
This is a hack, but I have "had" to do it with autofs on some of my
systems to get things to work.
On 4/12/07, Greenseid, Joseph M. <[EMAIL PROTECTED]> wrote:
The head node is not rebooting during this process. It is
simply up and running; a good example of this is a node installation.
On the head node, I am running the OSCAR Wizard; I am in
the monitor cluster screen, and the nodes are installing. Upon completion
of the install, I set them to reboot. They reboot (with the head node still
up, running the OSCAR Wizard), and no NFS file systems. When I click the
"complete cluster setup" button after they reboot, it says success, and then
do the "test cluster setup," the new nodes do not have /home mounted, and
that test fails...
--Joe
________________________________
From: [EMAIL PROTECTED] on behalf
of Michael Edwards
Sent: Thu 4/12/2007 2:15 PM
To: oscar-users@lists.sourceforge.net
Subject: Re: [Oscar-users] NFS file systems not mounting on
boot
Are you completely booting the head node before you boot
the client
nodes? If you don't the client nodes boot much faster than
the head
(because they run so few services) and they will generaly
finish
booting before the nfs server is up, so the mounts fail.
On 4/12/07, Greenseid, Joseph M. < [EMAIL PROTECTED]>
wrote:
> I am installing OSCAR 4.2.1 on an IA64 cluster. When the
nodes reboot (after installation, and also every time after that), they fail
to mount any NFS file systems. However, once the node has finished booting,
if I do a "mount /home" it mounts instantly.
>
> My exports file on my head node says:
>
> ~]$ cat /etc/exports
> #
> /home 10.2.148.1/255.255.255.0(async,rw,no_root_squash)
> /share 10.2.148.1/255.255.255.0(async,rw)
> ~]$
>
> My fstab entry on a compute node looks like this:
>
> # This file is edited by fstab-sync - see 'man
fstab-sync' for details
> /dev/sda3 swap swap defaults 0 0
> /dev/sda2 / ext3 defaults 1 2
> /dev/sda4 /tmp ext3 defaults 1 2
> /dev/sda1 /boot/efi
vfat defaults 1 2
> /dev/fd0 /mnt/floppy
auto noauto,owner 0 0
> none /dev/pts devpts defaults 0 0
> none /proc proc defaults 0 0
> nfs_oscar:/share /share nfs rw 0 0
> nfs_oscar:/home /home nfs rw 0 0
> none /dev/shm tmpfs defaults 0 0
>
>
> I tried changing nfs_oscar to the head node's eth0 IP
addr, and the file looks like this:
>
>
> # This file is edited by fstab-sync - see 'man
fstab-sync' for details
> /dev/sda3 swap swap defaults 0 0
> /dev/sda2 / ext3 defaults 1 2
> /dev/sda4 /tmp ext3 defaults 1 2
> /dev/sda1 /boot/efi
vfat defaults 1 2
> /dev/fd0 /mnt/floppy
auto noauto,owner 0 0
> none /dev/pts devpts defaults 0 0
> none /proc proc defaults 0 0
> 10.2.148.1:/share /share nfs rw 0 0
> 10.2.148.1:/home /home nfs rw 0 0
> none /dev/shm tmpfs defaults 0 0
>
> However, NFS mounting during boot fails in both cases.
>
> I get variations of the error message in my
/var/log/messages log from the failures during boot:
>
> Apr 12 13:23:58 compute-15-01 mount: mount: mount to NFS
server 'nfs_oscar' failed:
> Apr 12 13:23:58 compute-15-01 mount: System Error: No
route to host.
>
> or
>
> Apr 12 13:24:07 compute-15-01 mount: mount: mount to NFS
server 'nfs_oscar' failed: System Error: Connection refused
>
> or
>
> Apr 12 13:32:25 compute-15-01 mount: mount: mount to NFS
server '10.2.148.1' failed:
> Apr 12 13:32:25 compute-15-01 mount: System Error: No
route to host.
>
> As I said, in both instances (both IP address and
nfs_oscar hostname in the fstab), once booting was complete, the command
"mount /home" worked perfectly fine.
>
> Any idea why it isn't working during boot?
>
> Thanks,
> --Joe
>
>
-------------------------------------------------------------------------
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the
chance to share your
> opinions on IT & business topics through brief
surveys-and earn cash
>
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Oscar-users mailing list
> Oscar-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/oscar-users
>
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net 's Techsay panel and you'll get the
chance to share your
opinions on IT & business topics through brief surveys-and
earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV<
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV>
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net <mailto:
Oscar-users@lists.sourceforge.net>
https://lists.sourceforge.net/lists/listinfo/oscar-users
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the
chance to share your
opinions on IT & business topics through brief surveys-and
earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to
share your
opinions on IT & business topics through brief surveys-and earn
cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Oscar-users mailing list
Oscar-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/oscar-users
