EMC - Celerra Fileserver running out of inodes !!

[KapilRaj]

Hi,
I am not too sure if I can use this forum for this doubt ... if not please excuse me.

I was copying an online JFS of 2 GB on to one filesystem on this celerra box and it failed in between. I found 100 % usage of inodes .. Can I increase it ?. If yes how ??.

Cheers,

Kaps..

[Stuart Whitby]

I *think* that you can resolve this by copying the fs across in smaller chunks. Celerra is based on Windows, so there should be no reference to inodes *once the data is migrated*. However, during migration, an "offline inode" cache exists. My guess is that this is the one which is filling up, and that once your data has been migrated, you won't have a problem any more.

That's my theory, anyway. Hope it helps.

Stuart.

[Alzhy]

You did not mention how you are copying.

You might want to use vxdump/vxrestore ie:

vxdump 0f - /JFS_FILESYSTEM|(cd /Celerra_Dir;vxrestore rf -)

Hope this will fix your out of inodes situation. Note that this will only work if your source is a whole filesystem (or a mount point) -- if not, you might as well look at upgrading to VxFS 3.5 (OJFS 3.5) where vxdumps of directories are now allowed instead of whole filesystems...

[KapilRaj]

Guys,

Thanks for your help. I found out the correct option for celerra "nas_fs" command. I can specify -o nbpi=4096 which will allocate one inode per 4096 bytes. Now I am not able to confirm whether my change has done the trick or no...

I am copying by a tar

"cd ;tar cvf - . |ksh ("cd ;tar xvf -)"

nas_fs -i -s does not say what is the nbpi value is !!

Any idea ?

Kaps

[Alzhy]

Try the vxdump/vextrestore approach first. I had a similar problem with a different brand NAS...