CI: pkg/executor/importer:importer_test shard 29 of 38 frequently times out (flaky)
- Dominant language
- Go
- Stars
- 40.5k
- Forks
- 6.2k
- PR merge metrics
- PR metrics pending
Description
## Bug Report
### 1. Minimal reproduce step (Required)
Trigger the `ghpr_unit_test` Jenkins CI job on any PR. Observed on PR #66194 at:
https://do.pingcap.net/jenkins/job/pingcap/job/tidb/job/ghpr_unit_test/54792/
The `pkg/executor/importer:importer_test` Bazel test target, when run with 38 shards, exhibits a timeout on shard 29:
```
//pkg/executor/importer:importer_test shard 29 of 38 TIMEOUT in 60.4s
```
The target already carries `flaky = True` and `timeout = "long"` in its `BUILD.bazel`, indicating this is a known-intermittent test.
### 2. What did you expect to see? (Required)
All shards of `pkg/executor/importer:importer_test` should complete within the allotted timeout without flaky failures.
### 3. What did you see instead (Required)
Shard 29 of 38 consistently times out (~60 s) in CI runs, causing the overall `ghpr_unit_test` job to fail. The failure is unrelated to changes in the triggering PR; `pkg/store/helper:helper_test` (the PR's actual target) passed all 7 shards in the same run.
### 4. What is your TiDB version? (Required)
Branch: `master`
The `BUILD.bazel` for `pkg/executor/importer` already has `flaky = True`, but shard 29 appears to time out frequently enough to block PRs. Investigation should determine which test function in shard 29 is responsible for the excess runtime.
**Suggested labels:** `type/bug`, `component/lightning`, `severity/minor`
Contributor guide
Assessment
This issue has not been assessed yet.