IGNITE-28869 Fix IgniteWalRebalanceTest.testRebalanceCancelOnSupplyError

[wernerdv]

Thank you for submitting the pull request to the Apache Ignite.

In order to streamline the review of the contribution
we ask you to ensure the following steps have been taken:

The Contribution Checklist

• There is a single JIRA ticket related to the pull request.
• The web-link to the pull request is attached to the JIRA ticket.
• The JIRA ticket has the Patch Available state.
• The pull request body describes changes that have been made.
  The description explains WHAT and WHY was made instead of HOW.
• The pull request title is treated as the final commit message.
  The following pattern must be used: IGNITE-XXXX Change summary where XXXX - number of JIRA issue.
• A reviewer has been mentioned through the JIRA comments
  (see the Maintainers list)
• The pull request has been checked by the Teamcity Bot and
  the green visa attached to the JIRA ticket (see tab PR Check at TC.Bot - Instance 1 or TC.Bot - Instance 2)

Notes

• How to Contribute
• Coding abbreviation rules
• Coding Guidelines
• Apache Ignite Teamcity Bot

If you need any help, please email dev@ignite.apache.org or ask anу advice on http://asf.slack.com #ignite channel.

[anton-vinogradov]

Reviewed both the diagnosis and the fix — the root cause and the approach look correct. I traced it end to end:

• A cache version's topVer field is readyTopologyVersion + offset (GridCacheVersionManager.next(topVer)), where offset = (gridStartTime - TOP_VER_BASE_TIME) / 1000 (seconds), captured in onLocalJoin.
• GridCacheVersion.compareTo compares topologyVersion() first, so offset dominates the reset readyTopologyVersion.
• The failing assertion is ATOMIC_VER_COMPARATOR.compare(entry.ver, newVer) <= 0 in GridCacheMapEntry (Invalid version for inner update), i.e. the new write's version must be >= the recovered entry's version.

On a fast full-cluster restart gridStartTime advances by less than the topology-version reset, so the new writes get a strictly lower topVer and trip the assertion. The pause is placed correctly — before startGrid(0), since offset is captured at node-0 join — and makes the new offset exceed the old one by ~5, which comfortably covers the ~2 topology-version gap (3-node+activate vs 1-node+activate). cleanPersistenceDir() runs per test, so the gap is intra-test only and the margin is large; the fix is deterministic (wall clock only moves forward), not a new source of flakiness. LGTM.

One optional nit: the 5s constant is implicitly tied to the topology this test builds. It's fine today, but worth spelling out the rationale so a future change to the started-grid count doesn't silently shrink the margin — inline suggestion below.

Also worth attaching the TC Bot green visa to the ticket before merge.

[anton-vinogradov]

Now that the PR carries the deterministic version, two comment-only nits on the newly added comments (no code changes, both optional).

[ignitetcbot]

TCBot Test Analysis

• TeamCity: --> Run :: All Results
• Branch: pull/13323/head
• Base: master
• Tested commit: 82e2785

Possible Blockers (0)

No blockers found.

New Tests (0)

No new tests found.