I’ve added a file-based caching layer with settings-based TTL to the volumes() function, but that cache was only available to the worker processes. I’ve changed it to accept cache if the mountpoints match what’s currently inn
I just realized that if I add a configurable cache to mountpoints(), update it to include mount points so a bigger value (say, 24h) will let HDDs idle.
To minimize hits to df (which may wake drives), set volumeMetadataTtlMs=0 and mountpointsTtlMs=0 which will cache volume metadata indefinitely.