You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@seatunnel.apache.org by "davidzollo (via GitHub)" <gi...@apache.org> on 2026/08/04 02:22:01 UTC
[PR] [Test][E2E] Give PostgresCDCIT post-restore assertions their own budget [seatunnel]
davidzollo opened a new pull request, #11644:
URL: https://github.com/apache/seatunnel/pull/11644
## Purpose
`PostgresCDCIT#testPostgresCdcSnapshotOnlyAndCommittedOffsetStartupModes` fails intermittently in `all-connectors-it-3` and `updated-modules-integration-test-part-3`:
```
PostgresCDCIT.testPostgresCdcSnapshotOnlyAndCommittedOffsetStartupModes:635 » ConditionTimeout
Assertion condition ... expected: <1> but was: <0> within 1 minutes.
```
Seen on #11597 and #11569 within hours of each other. Neither PR touches Postgres CDC — one adds a Zeta slot-usage API, the other propagates JDBC XA commit failures.
## Root Cause
This class has ~23 `await()` assertions. Twenty of them wait on a CDC round trip against an already-running job, and 60s is comfortable for that.
Three of them — at lines 635, 920 and 1022 — run immediately after `savepointJob` + an async `restoreJob`. Those must additionally absorb the entire restore (cluster restart, connector re-initialization, replication slot reattach) before the round trip they assert on can even begin, yet they were given the same 60s. On a loaded runner the restore alone can consume the budget, so the assertion times out having never had a chance to observe the row.
## Changes
Introduce `RESTORE_ASSERT_TIMEOUT_MILLIS = 180000` and use it for **only those three** post-restore waits. The other ~20 assertions keep their 60s budget.
This mirrors the existing `DEBEZIUM_JSON_RECORD_WAIT_TIMEOUT_SECONDS = 180L` precedent already in this class for waits that absorb extra latency.
No assertion is weakened — every condition asserted, and the values asserted, are unchanged. Only the three waits that have strictly more work to absorb get a budget that reflects it.
## Does this PR introduce _any_ user-facing change?
No. E2E test only.
## How was this patch tested?
The Postgres CDC E2E jobs on this PR exercise all three changed waits.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] [Test][E2E] Give PostgresCDCIT post-restore assertions their own budget [seatunnel]
Posted by "zhangshenghang (via GitHub)" <gi...@apache.org>.
zhangshenghang merged PR #11644:
URL: https://github.com/apache/seatunnel/pull/11644
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org