[OpenAFS] running out of inodes on OpenAFS file-server
Fri, 10 Jul 2009 16:25:41 +0200
Vladimir Konrad wrote (Fri Jul 10 2009 15:53:44 GMT+0200 (CEST))
> One of my servers is running out of inodes on /vicep?? partition...
> Is it safe to:
> 1. shut down the OpenAFS file-server in question.
> 2. copy the content of /vicep?? partition somewhere safe
> 3. re-create ext3 file-system on the affected partition
> with better settings (for inodes)
> 4. copy the content of /vicep?? partition back
> 5. start the fileserver
> I vaguely remember that the file-server needs inode numbers to get to the files on /vicep??,
> and these would obviously change when doing the above. Or the inode numbers are scanned on
> file-server start-up (so the above would not break anything)?
This would only be the case if you were running a fileserver with inode
partitions, which you hopefully are not. Does /vicep*/AFSIDat exist?
Then you have namei partitions, which should allow for such stunts.
> Other option comes to mind: move all the volumes (stored on the partition)
> to a different /vicep?? partition (using vos move), shut down the file-server,
> re-create the ext3 file-system with better settings, move the volumes back,
> start the file-server.
This will work. Alternatively (if the data volume is large), you might
prefer moving part of it away, then shrinking your ext3 partition and
adding another (effectively splitting your space).
Moving data back is only necessary if you want that data to be on that
particular server, too.
> This is on Debian GNU Linux, with OpenAFS file-server 1.4.2 .
Is that the newest one available in your Debian? Upgrading fileservers
is not dangerous, so I'd recommend it.