*: allow UnionBlockInputStream ignore the blocks#3638
*: allow UnionBlockInputStream ignore the blocks#3638ti-chi-bot merged 25 commits intopingcap:masterfrom
Conversation
|
[REVIEW NOTIFICATION] This pull request has been approved by:
To complete the pull request process, please ask the reviewers in the list to review by filling The full list of commands accepted by this bot can be found here. DetailsReviewer can indicate their review by submitting an approval review. |
|
/run-all-tests |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/350/cobertura/ lines: 41.9% (46383 out of 110695) |
|
/run-all-tests |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/356/cobertura/ lines: 41.9% (46383 out of 110699) |
dbms/src/Flash/Mpp/MPPTask.cpp
Outdated
| FAIL_POINT_TRIGGER_EXCEPTION(FailPoints::exception_during_mpp_non_root_task_run); | ||
| } | ||
| } | ||
| FAIL_POINT_PAUSE(FailPoints::hang_in_execution); |
There was a problem hiding this comment.
The failpoint here seems useless because once reach here, the execution is finished.
There was a problem hiding this comment.
I fail to find a better place. If the failpoint is triggered in ExchangeSender, multiple exceptions will be thrown and the error message will be mismatch.
|
/run-all-tests |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/396/cobertura/ lines: 42.3% (46961 out of 111016) |
|
/run-all-tests |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/401/cobertura/ lines: 42.4% (47048 out of 111015) |
|
/run-all-tests |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/408/cobertura/ lines: 42.4% (47051 out of 111014) |
|
/run-integration-test |
| template <class StreamWriterPtr> | ||
| void StreamingDAGResponseWriter<StreamWriterPtr>::write(const Block & block) | ||
| { | ||
| FAIL_POINT_PAUSE(FailPoints::hang_in_execution); |
There was a problem hiding this comment.
I think its better to put these failpoint to ExchangeSender
dbms/src/Flash/Mpp/MPPTask.cpp
Outdated
There was a problem hiding this comment.
Are you sure tracing will be GA in the next release, because if not then we will have to release a version without this information.
|
/run-integration-test |
|
/merge |
|
@fuzhe1989: It seems you want to merge this PR, I will help you trigger all the tests: /run-all-tests You only need to trigger If you have any questions about the PR merge process, please refer to pr process. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
|
This pull request has been accepted and is ready to merge. DetailsCommit hash: d05bcbe |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/429/cobertura/ lines: 42.5% (47528 out of 111811) |
|
Coverage detail: https://ci-internal.pingcap.net/job/tics_ghpr_unit_test/430/cobertura/ lines: 42.5% (47528 out of 111811) |
What problem does this PR solve?
Issue Number: close #xxx
Problem Summary:
HashJoinBuildInputStream and ExchangeSender will return blocks through UnionBlockInputStream and all blocks will first contend for being pushed into a ConcurrentBoundedQueue. This can be a performance bottleneck:
What is changed and how it works?
Introduce a new template parameter
ignore_blocktoUnionBlockInputStreamthat ignores blocks atonBlock.To not break the test, also move some fail points from
MPPTask::runImpltoStreamingDAGResponseWriter.Check List
Tests
Side effects
Documentation
Release note