2012-04-27 05:19 PM
I have a question about Inode allocation
We just ran into an issue where the volume ran out of Inodes. I increased the Inodes for the volume, and this looks to have taken care of the issue for now.
My concern is that we may likely run into this again. I was wondering if anyone out there knows if there is such a thing as Dynamic Inode Allocation for Netapp volumes?
I haven't been able to find any literature on this, but thought I would also try the good folks in the Communities to see if there may be
Thanks so much for any advice
2012-04-28 02:43 AM
I dont know that you can do that dynamically, maybe a NTAP guy can answer? However there are a couple of auto features that might be of interest:
1a. If you use snapshots in that volume you could consider using less snapshots and also use snap autodelete. Deleting snapshots should free up inodes. I havent tried what I am mentioning here, just putting it out there. One of the challenges with this is that once a snapshot has been deleted, depending on various factors as workload etc, it can take some time to get the freed space back.
1b. Use less snapshots as mentioned, without snap autodelete should also free up inodes. This would need to be tuned and if implemented properly will free up
space ahead of you getting into problems.
2. The other way to add inodes is to auto grow the volumes, but again, I m not sure if the volume will ONLY grow when its run out of space, which is typically does. Or if it will autogrown once it runs out of inodes. From experience I have only seem them autogrown when they run out of space. This would add inodes straight
Hope this helps, I reckon it ll draw some debate for starters, that can always be helpful
2012-04-28 07:57 AM
No dynamic allocation but you could alert on it. Whatever threshold in df -i. Then maybe run a job To increase maxfiles. Default is 32k and can be increased / down to 4k.
It would be nice to have this as an autograph policy like vol autogrow. Then also the same for maxdirsize which we also run into.