Fix
-
GitHub #135: drop the full-NAS snapshot scan from
volume_snapshot_info. The earlier implementation calledpool.snapshot.querywith no filter and noextra.properties, so on a NAS with several thousand snapshots it fetched every snapshot on the array (~20 MB JSON, ~33 s per call) and then filtered client-side.qm rollbackmakes three of those calls per invocation, which put rollbacks in the multi-minute range for anyone with a busy TrueNAS.volume_snapshot_infonow delegates to the existing_tn_snapshot_query_datasetshelper, which asks the middleware to filter by dataset and return only thecreationproperty. On the reporter's NAS the per-call time dropped from ~33 s to ~1.7 s; on our test cluster the call completes in ~10 ms. Also drops the unused_tn_snapshotshelper (dead code, zero callers).
Validation
perl -cclean on the cluster node.- Behavioural smoke on a volume with one snapshot: 10 ms, correct name, correct creation timestamp.
The .deb artifact is uploaded by the tag-push CI.
Full Changelog: v2.1.24-beta5...v2.1.24-beta6