/nsr filesystem is full.

[kl11]

I inherited a backup server that does backups to disk. The problem
is that the /nsr directory is located on the same ZFS pool as the
backups, so when the space ran out for the backups, the /nsr directory
also ran out of space.

All the networker processes are unresponsive. I tried a
nsr_shutdown -f , but that doesn't do anything. I have tried
manually killing nsrd, ie kill <nsrd_pid>, but that doesn't work.
At this point I will probably have to reboot the server.

I have some free space on another partition, where I can
move some of the indexes. How much space is needed in the
/nsr directory?

What do I need to order get network running again, so I
can clean up some of the disk space?

I'm not overly familiar with networker. I'm running
Solaris 10 update 3 with networker 7.3,REV=364.

File system:
tank 16T 87G 250M 100% /nsr
tank/disk1 16T 6.6T 250M 100% /nsr/disk1
tank/disk2 16T 7.2T 250M 100% /nsr/disk2
tank/disk3 16T 2.1T 250M 100% /nsr/disk3

Thanks,
Kevin

 

[605]

That is almost a worst case scenario. Compared to your backup data, the index space is negligible - so you can not avoid to migrate the backup data.

First i would do is to stop (kill -9) all nsr* processes. If you really must reboot the server, make sure that NW will not start them later on.

Then i would delete /nsr/logs/daemon.log (don't remember whether it was already daemon.raw) as well as /nsr/logs/messages and /nsr/logs/savegrp.log. This should give you enough disk space to restart NW.

Now add at least one new file system.

Start NW.
Then set the current AFTDs to read only.
Create some new AFTDs, label their media (don't know whether migrating to the same pool was already supported) and migrate your backups using a nsrstage command/script.
I am pretty sure this will take longer.
Proceed with the next AFTDs until all media has been migrated.

The tasks themselves are pretty easy. However, if you are not familiar with staging, you might also loose data. If in doubt, get help from a NW professional.