Files
mxaccessgw/docs/deploy/2026-11-12-pending-worker-time-fixes-226-228.md
T
Joseph Doherty 31f7f9fc6e
ci / nightly-windev (push) Has been skipped
ci / windows-x86 (push) Successful in 2m19s
ci / portable (push) Failing after 24m22s
ci / java (push) Failing after 24m39s
docs(deploy): DEV vd03 on 76a71bc8, Worker time fixes #226 + #228 verified live (#229)
Server and x86 Worker from 76a71bc8 on wonder-app-vd03, about 9 s downtime, env block and
pepper prefix unchanged. Write2 KeepsSentTimestamp passes; RuntimeAlert reads
2026-07-28T20:02:46.830Z. QA still pending (#170). Upgrade-script gaps filed as #230.

Claude-Session: https://claude.ai/code/session_01GT67ACmpMR7b9LGRPLijMi
2026-11-12 21:23:40 -05:00

5.3 KiB

Pending: Worker time-conversion fixes #226 + #228 (DEV first, then QA)

Status: DEV deployed 2026-11-12 (76a71bc8, both checks passed, record 2026-11-12-vd03-76a71bc8-worker-time-fixes.md); QA NOT deployed (#229, gated by #170). This is the deploy note for the next gateway deploy. Ship both fixes in one deploy: they are both in the x86 Worker, both change timestamps consumers see, and both are on main.

Fix Commit Behaviour change on the box
#226 wnwrap alarm DATE+TIME are already UTC; stop adding GMTOFFSET-DSTADJUST 892bdc1a Every wnwrap alarm time (QueryActiveAlarms, alarm events) moves 4-5 h earlier, to correct UTC. OtOpcUa/ScadaBridge dedup keyed on raise time sees each existing alarm once with the new time.
#228 Write2/WriteSecured2 (+ bulk) time argument sent as box-local time 9ef6fc84 The stored source timestamp of a timestamped write equals the instant sent, instead of being shifted by the box's UTC offset (+5 h on a UTC-5 box in winter, +4 h in summer). No consumer calls Write2 today (ScadaBridge and OtOpcUa have no call site).

Build Server and Worker together from one main SHA at or after 9ef6fc84 (the upgrade script refuses staged builds from different SHAs; they share WorkerProtocolVersion). Procedure, traps and rollback: docs/runbooks/DeployUpgradeNssmService.md and scripts/deploy/Upgrade-MxGatewayService.ps1.

What else rides along

  • DEV vd03 runs 2b13564e. 2b13564e..9ef6fc84 adds only #226, #227 (opt-in emit_stream_attached, additive proto, Contracts/Client 0.4.0) and #228, plus docs/tests. Nothing else changes for current consumers.
  • QA vq03/vq04 run 4f931e4. A main build also carries the 2026-10-21 review remediation and the operator-authenticated alarm ack (488649c). The QA deploy is therefore gated by #170: every QA OtOpcUa node with a Galaxy driver must first run lmxopcua dc0a24f2+ and have an AlarmAck fallback account (or per-key ack identities) configured, or every Galaxy ack is refused. Close that gate first or in the same window.

DEV (wonder-app-vd03) first

  1. Build on windev in a scratch folder from the chosen SHA: Worker dotnet build src\ZB.MOM.WW.MxGateway.Worker\ZB.MOM.WW.MxGateway.Worker.csproj -c Release -p:Platform=x86 and run Worker.Tests x86 (expect 0 failures; 783 passed / 12 skipped at 9ef6fc84); publish the Server self-contained as in runbook section 4. Do not touch windev's own MxAccessGw service.
  2. Stage both to vd03 over SMB, run the upgrade script on the box (deploy activity, WinRM in scope): Upgrade-MxGatewayService.ps1 -ServerSource <stage>\Server -WorkerSource <stage>\Worker -ExpectedSha <sha> -Label issue226-228. Single box: stop, swap, start once (trap 1, OrphanWorkerTerminator).
  3. Runbook section 7 checks (health 200, pre-upgrade mxgw_ key authenticates, wrong key rejected, Worker path proven by a read).
  4. #228 check, from the Mac, no remote exec: with a scoped key for MESReceiver_025.* (Z28062, never Z28061), run MXGATEWAY_LIVE_GATEWAY_ASSERT_WRITE2_TIMESTAMP=1 dotnet test clients/dotnet/ZB.MOM.WW.MxGateway.Client.Tests --filter FullyQualifiedName~LiveGatewayWrite2Tests with MXGATEWAY_LIVE_GATEWAY_ENDPOINT, MXGATEWAY_LIVE_GATEWAY_API_KEY_FILE and MXGATEWAY_LIVE_GATEWAY_WRITE_ITEM=MESReceiver_025.MoveInBatchID (docs/GatewayTesting.md, "Live gateway write"). Both tests must pass: Write2_SupervisoryAdvise_KeepsSentTimestamp failed on 2b13564e with sent 01:36:51Z / stored 06:36:51Z. Revoke and delete the key afterwards.
  5. #226 check: QueryActiveAlarms for Galaxy!CVDAisle_1.TableAlarms_008.RuntimeAlert reads 2026-07-28T20:02:46.830Z (was 01:02:46.830Z on 07-29), and a fresh alarm's time equals Runtime.dbo.Events.EventTimeUtc on wonder-sql-vd03 (read over SQL, not WinRM).
  6. Write the deploy record in docs/deploy/ and update the scadaproj index row.

QA (wonder-app-vq03 / vq04) after DEV passes

  1. Confirm the #170 prerequisites above on both QA OtOpcUa nodes.
  2. Same SHA and staged builds as DEV. The gateways are singletons per box: deploy vq03, run the section 7 checks, then vq04 (rolling). The vq04 gateway depends on vq03 for Galaxy SQL and alarms.
  3. #228 check as DEV step 4 against each QA gateway, with a Z28062 test attribute that exists on QA (pick and record it; MESReceiver_025.MoveInBatchID is the DEV item). #226 check as DEV step 5 against the QA Historian.
  4. Deploy record, index row.

Rollback

Runbook section 8: stop, confirm no Server/Worker process, rename the Server.bak-*/Worker.bak-* pair back, start once. Both fixes are Worker-only behaviour; no store or schema changes, so rollback restores the old (wrong) times and nothing else.

Evidence the fix works (windev, 2026-11-12, Eastern Standard Time, UTC-5)

The live smoke WorkerLiveMxAccessSmokeTests.GatewaySession_WithLiveWorker_Write2ValueWithTimestampToAdvisedItem (in-process gateway, its own worker; the MxAccessGw service was untouched) on TestChildObject.TestInt:

Worker Sent Stored Result
9ef6fc84 (fix) 2026-11-13T01:46:31Z 2026-11-13T01:46:31Z Passed
9ef6fc84 with the conversion removed (negative control) 2026-11-13T01:47:32Z 2026-11-13T06:47:32Z Failed by exactly +5 h

Both runs got OnWriteComplete Ok and restored the original value.