I wanted to get confirmation on how space is managed on a mixed local and remote index cluster.
I know that maxVolumeDataSizeMB is ignored on remote/s3 enabled indexes, and that eviction_padding in server.conf can control how the cachemanager will start evicting from local cache on the indexers.
maxvolumedatasizeMB is ignored and can lead to disk filling up unexpectedly. Depending on version of splunk, it will either start evicting evictable buckets at the eviction_padding threshold or it wont evict and then eventually pause indexing when it hits minfreespace.
Advice is not to have s2 indexes and non-s2 indexes in the same volume with maxvolumedatasizeMB as it's unlikely either of the above behaviors are desired. They should keep non-s2 and s2 indexes in separate volumes.
How about the indexes that were migrated to SmartStore? The homePath remains the same after migration and it refers to a volume that had maxVolumeDataSizeMB.