I have two datastores for PBS: a 4x2 raid10 (ZFS) and a ZFS mirror. GC for the latter works just fine. For the former, every time it runs, in phase 1 (marking), I see:
found (and marked) 602 index files outside of expected directory scheme. (602 is the total number of chunks). Notwithstanding the above, backups still seem to work, and I've successfully restore VMs from the raid10 datastore. Any clues as to how serious this is would be much appreciated.
I suspect this has to do with namespaces. The raid10 has 3 (PBS-Daily, PBS-Weekly and PBS-Monthly), yet the .chunks directory is at the parent level (presumably so deduplication will work better). I'm guessing that check is a bug in the PBS software?
found (and marked) 602 index files outside of expected directory scheme. (602 is the total number of chunks). Notwithstanding the above, backups still seem to work, and I've successfully restore VMs from the raid10 datastore. Any clues as to how serious this is would be much appreciated.
I suspect this has to do with namespaces. The raid10 has 3 (PBS-Daily, PBS-Weekly and PBS-Monthly), yet the .chunks directory is at the parent level (presumably so deduplication will work better). I'm guessing that check is a bug in the PBS software?
