Skip to content

armgit / Pulse

Public Imported from GitHub
0

Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.

Compose truthful TrueNAS snapshot times with current recovery contracts

pulse-triage[bot]4058253357e4
diff --cc docs/release-control/v6/internal/subsystems/storage-recovery.md
index 31d5a0396c,35fdd41cca..697c57035e
--- a/docs/release-control/v6/internal/subsystems/storage-recovery.md
+++ b/docs/release-control/v6/internal/subsystems/storage-recovery.md
@@@ -21,22 -21,25 +21,41 @@@
## Purpose
+### Saved configuration is not successful live recovery
+
+Configuration import returns an explicit saved-but-incomplete error and failed
+audit outcome when an attempted live alert, notification or guest metadata
+reload fails. Independent reloads continue, but the recovery dialog must not
+close and claim success or schedule its normal success reload. The response
+names only fixed components, never private errors, destinations or archive
+inputs, and fits the consuming API client's plain-text bound. Persistence is
+already committed: this is not rollback, nor a reason to repeat the import.
+The existing bounded replay admission protects newer edits after partial apply.
+Initialization without a monitor remains supported; absent sub-managers in a
+present monitor fail without panic. The real-import runtime-outcome and router
+contract tests, frontend transport suite and source-bound consuming browser
+proof cover the distinction. No workload, database schema, guest-agent request
+or native recovery operation changes.
+
+ ### TrueNAS snapshot collection is not creation evidence
+
+ Undated or zero-date ZFS snapshots retain no start/completion time and have
+ Unknown outcome. Their IDs remain stable across polls regardless of collection
+ time; native dated snapshot IDs/times and replication semantics are unchanged.
+ On store open, former successful snapshot rows with no start become undated
+ Unknown observations: completion-only rows used the old collection-time fallback,
+ and rows with neither date could come from a zero native date. Every
+ row, ID and reference remains; valid detail objects retain the old synthetic
+ value as `legacyCollectedAtMs`, not creation evidence. No schema or retention
+ change, deletion, native snapshot, replication or restore is performed.
+
+ Undated snapshots cannot create successful-point, attempt-date or dated-series
+ claims from collection/insertion time. The mapper, reopened-store, rollup and
+ series controls in `mapper/truenas/mapper_test.go` and `store/store_test.go`
+ cover missing/zero/known dates, repeated collection, legacy correction and
+ provider/kind/mode isolation. The consuming TrueNAS Protection table and guide
+ retain missing dates as unavailable evidence rather than a fresh backup.
+
### Future backup dates cannot establish current protection
Protection freshness and verification must fall at or before their evaluation