#Setting inode limit for volumes
1 messages · Page 1 of 1 (latest)
you need to do that on the cluster/svm directly. volume modify ... -files ...
Thanks for the answer. 😕
I found this KB article that suggests using using flexgroup or ontap-san drivers to alleviate the issue. Is that generally an option that can be used, or are those completely different products compared to ontap-nas?
from a user point of view, it is still a PVC.
however, here comes a few questions:
- is RWX required? (with Block, not that common)
- how many PVCs do you have? (building a FlexGroup consumes more ONTAP FlexVol that ONTAP-NAS)
The ontap-san driver creates iSCSILUNs instead of NFS volumes, so your worker nodes have to create their own filesystem on those LUNs. There might be file systems that support more files per default, but even there you'd need to check that you don't run into inode limits.
The ontap-flexgroup driver creates FlexGroups that can have up to billions of files, but they create many more volumes (called "constituents") in the backend for each trident volume (4 to 20), and depending on the number of trident volumes you intend to create, you will run into maximum-volume-count issues on your cluster at some point