mirror of
https://github.com/rcourtman/Pulse.git
synced 2026-10-24 04:33:17 +00:00
rcourtman
41cea2ffab
## Problem The Proxmox disk poller keeps evidence it could not collect this poll by looking up the previous poll's record of the disk: by serial, then WWN, then the Proxmox source ID (`instance-node-devpath`), then the device token. The source ID and device token name a slot, not a disk, and a replacement in the same slot keeps both. So: - a replacement reporting a different serial or WWN took the old disk's retained temperature, controller, pool and I/O; - a replacement record with an empty serial also took the old disk's serial. The registry keys disks on serial, so the new disk became the old disk's canonical resource and SMART metric series, and every later poll carried the borrowed serial forward. Current Proxmox reports a missing serial as the literal `unknown`, which the poller records as reported, so the serial copy needs a producer that sends an empty serial, such as the host-agent fallback rows. The evidence copy happens with real Proxmox data. ## Change - `previousPhysicalDiskEvidence` matches by hardware identity through `diskinventory.HardwareIdentityMatch` (serial first, then WWN, each against both fields of the earlier record), and refuses a slot match when both records carry a serial or WWN and none agrees. Missing identity is still not a conflict: placeholders such as `unknown`, and an empty serial, keep the slot's evidence as before. - The registry's merged disk view keeps the host agent's WWN, which the agent spells as unpadded NAA 5 fields (`5-c50-a1b2c3d4`, `internal/hostagent/smartctl.go` `formatWWN`) while Proxmox reports `0x5000c500a1b2c3d4`. `HardwareIdentityMatch` now re-pads that spelling for WWN fields only, so one disk seen by both reporters is not taken for a replacement. The registry's linked-disk join uses the same comparison. ## Tests - `TestPhysicalDiskReplacementInSameSlotDoesNotInheritPreviousSerial` drives the poller and registry across two polls after a swap; on base the replacement inherits the old serial and canonical resource. Its controls keep a serial-omitting same disk on its resource. - `TestPreviousPhysicalDiskEvidenceRejectsSlotMatchAcrossConflictingIdentity` covers renamed devices matched only by identity, PVE vs agent WWN spellings, SAS address serials with a matching WWN, placeholders, and conflicting serial/WWN by source ID and by device token. - Negative controls via `go test -overlay`: on base the replacement, conflict and renamed-device cases fail; with the guard but without the WWN re-padding, the same-disk PVE/agent WWN cases fail. - `pkg/diskinventory`, `internal/unifiedresources` and `internal/monitoring` pass in full. Follow-ups, not in this change: the host-agent SMART merge's device-path fallback can still fill or promote a serial from a stale SMART row; QEMU default serials (`drive-scsi0`) collapse nested Proxmox nodes' disks into one resource; and the registry merges disks across parents on a canonical-ID collision.