Ignite backup problem ?

[rveri]

Hi All ,

Found Ignite backup failed, from the logfile.
While re-running getting this error message continously :

"NFS server srv-abc not responding still trying."

details:

* bdf hangs

* bdf -l : works fine ,all local filesystems shows.

* tried to ping the server srv-abc , its pingable.

* found someone earlier mounted one nfs filesystem , from srv-abc server : ( using this command : # mount -F nfs srv-abc:/mnt/fs1 /remotefs ) , and now this filesystem is no longer available , causing bdf to hang. Even # cd /remotefs hangs.



Whats wrong. What can be done , to unmount that filesystem.

I have tried :

# umount /remotefs
nfs umount: nfs_unmount: /remotefs: is busy
umount: return error 1.


Pls suggest.

Thanks ,

[njia_1]

Hi

Just like what you said, the remote file system is not longer available and that is why bdf hangs.

Have you tried to stop/restart nfs.server and nfs.client ?

Are you doing an ignite backup cross the network ?

Have a look at ignite's man page, there may be an option to by pass that directory.

thanks

[Devender Khatana]

Hi,

use fuser to kill the processes using files in this file system if any.

#fuser -ck /remotefs

The try unmounting it. If do not succed then if there are no other NFS file systems exported/imported from this server do

#/sbin/init.d/nfs.client stop

#/sbin/init.d/nfs.server stop

#/sbin/init.d/nfs.core stop

If the issue gets resolved by this then restart the services again.

HTH,
Devender

[Devender Khatana]

Hi Again,

The earlier post actually concerned about the bdf problem only. So for the ignite problem I think you should specify to backup only vg00 using -x switch.

Still it may give problem as it will use bdf and fstab to find out the currently unmounted file systems without which it can not proceed.

HTH,
Devender

[rveri]

Devender ,

This is the command I am using for Ignite backup :

* /opt/ignite/bin/make_tape_recovery -A -v -a /dev/rmt/0mn -x inc_entire=vg00 -n 2

* Also this is a adoptive node of a 2 node cluster . Will be there any impact , if the nfs.client stop , nfs.server stop & nfs.core stop , executes.

Please suggest.

Thanks.

[Devender Khatana]

Hi,

If it is part of cluster, then I would recommend to reboot it as the other node will take care of applications. The impact may there if some of the packages require NFS services.

HTH,
Devender

[njia_1]

Hi there,

do a nfs.server stop/start and nfs.client stop/start on the adoptive node should not have any impact on the packages as long as the package is not running on the adoptive node.

do a showmount -e first on that node to see if any file system being exported.

thanks

[rveri]

Hi ,

showmount shows the following error:


# showmount -e
showmount: infsrv12: RPC: Program not registered.

What could be the problem.

Thanks.

[Chauhan Amit]

Check out these Links with similar problem,

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=986938

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=106047

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=938994

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=68070

-Amit

[Muthukumar_5]

It is a problem with *NFS*.

Some body is trying to use /remotefs.

Check as,

# fuser -k /remotefs

It will release that file system usage.

# cd /
# umount /remotefs

It is unmounted now.

Try to restart NFS as,

# /sbin/init.d/nfs.server stop
# /sbin/init.d/nfs.server start

# /sbin/init.d/nfs.core stop
# /sbin/init.d/nfs.core start

# /sbin/init.d/nfs.client stop
# /sbin/init.d/nfs.client start

Check rpcinfo status for debugging.

--
Muthu

[Muthukumar_5]

After using fuser -k /remotefs and umount /remotefs, Remove /etc/fsstab entry for /remotefs entry (If you have added already). Else it will be mounted upon reboot or mount -a.

--
Muthu

[rveri]

Hi Muthu ,and all...

* the nfs filesystem entry its not there in /etc/fstab
* strings /etc/mnttab shows the nfs entry ,
i.e

srv-abc:/mnt/fs1 /remotefs nfs rsize=32768,wsize=32468,NFSv3 0 0 1138033970

* I tried to rerun , after removing the nfs entry , from mnttab , and make_tape_recovery started fine..

* Now its re-run again , and its hanged in ..

" * Creating Archive File List " , (just waiting , dont know if it hanged or going on)



---
ps -ef | grep ignite shows , already two instance of (list_expander) ignite that hanged.

# ps -ef | grep ignite
root 21011 1 0 03:28:47 pts/tu 0:00 /opt/ignite/lbin/list_expander -s -S -f /var/opt/ignite/recover
root 13415 1 0 02:07:01 ? 0:01 /opt/ignite/lbin/list_expander -d -f /var/opt/ignite/recovery/l
root 29602 28514 0 05:07:13 pts/tv 0:00 sh -c /opt/ignite/lbin/list_expander -s -S -f /var/opt/ignite/r
root 4107 28434 0 05:57:41 pts/tx 0:00 grep ignite
root 28514 21091 0 05:06:48 pts/tv 0:00 /opt/ignite/bin/make_tape_recovery -A -v -a /dev/rmt/0mn -x inc
root 29604 29602 0 05:07:13 pts/tv 0:00 /opt/ignite/lbin/list_expander -s -S -f /var/opt/ignite/recover
---------------------------------------

- What list_expander do .., ?
- why it is hanged at "*Creating Archive File list " ,?
it is supposed to move to the next action , i.e
"* Archiving contents of server1 via tar image to local device/dev/rmt/0mn."


Any help /suggestion is appreciated.
Thanks.,

[rveri]

Also ,
this is a node of a two node cluster. Earlier thought no package is running on this but , cmviewcl shows ,one Package is running on this particular node.

thanks.

[Devender Khatana]

Hi,

It is creating a list of file in the file systems specified in the backup command. At this stage it may take few minutes and it will actually depend upon the system utiligation at present and the data for which the list has to be created. The list is created in file named flist and you can tail this file to see the progress. The file will be located in directory
/var/opt/ignite/recovery/latest

HTH,
Devender

[Devender Khatana]

Hi,

There should not be any difference in taking ignite backup even if packages are running.

Also you mentioned of removing entry from /etc/mnttab. How did you do that? The file is not supposed to be edited manually.

Look
man mnttab

HTH,
Devender