matrixorigin / matrixorigin/matrixone
ci: data branch merge timeout in restart BVT (e2e BVT Optimistic)
- Dominant language
- Go
- Stars
- 1.9k
- Forks
- 311
- Avg merge
- 1d 3h
- Merged PRs (30d)
- 768
Description
## Summary
PR [#24609](https://github.com/matrixorigin/matrixone/pull/24609) (add degrees) e2e BVT restart test fails: `data branch merge` in `merge_1.sql` times out after 300s when re-running BVT after MO restart.
## CI Run
- Job: https://github.com/matrixorigin/matrixone/actions/runs/26491501744/job/78010186996
- Step: "Start BVT Test for MO with Version of Head Restarted"
- Mode: Standalone LAUNCH Optimistic
## Failure Details
**First run** (before restart): merge_1.sql passes in 0.67s, all 81 cases success.
**After MO restart**, re-running merge_1.sql:
| Row | SQL | Error |
|-----|-----|-------|
| 14 | `data branch merge t2 into t1;` | **timeout** — MO does not return result in 300000ms |
| 16 | `select * from t1 order by a asc;` | failed (reconnect triggered) |
| 17 | `data branch diff t2 against t1;` | no such table merge_1.t2 |
| 19 | `drop table t1;` | no such table merge_1.t1 |
| 20 | `drop table t2;` | no such table merge_1.t2 |
Success rate: 93% (76/81 success, 4 failed, 1 abnormal), below 100% threshold.
Additionally, `sample.sql` also fails after reconnect (`create temporary table sample05`).
## Analysis
- `data branch merge` succeeded in the initial BVT run (0.67s)
- After MO restart, the same merge hangs/times out
- Subsequent failures are cascading — mo-tester reconnects after timeout, but merge state is corrupted
- Not a duplicate of [#24099](https://github.com/matrixorigin/matrixone/issues/24099) (multi-CN PESSIMISTIC `ErrTxnNeedRetryWithDefChanged`) or [#23934](https://github.com/matrixorigin/matrixone/issues/23934) (merge_2.sql dup entry)
- Likely related to MO internal state after restart affecting git4data branch merge operations
Contributor guide
Assessment
This issue has not been assessed yet.