Yuqi Du has posted comments on this change. ( http://gerrit.cloudera.org:8080/19463 )
Change subject: [tserver] keep the order of appling ops the same as raft wal ...................................................................... Patch Set 1: (10 comments) > Patch Set 1: > > (11 comments) > > Is this specific issue of out-of-order ops seen in any environment? Yes. The environment can get easily if provide enough qps of WriteRequest, I'll provide a test for this. > It would really help if you could also capture one liners about following > points: > 1. How can out-of-order ops happen given the current concurrent thread-pool > implementation? I have described that, and it's too simple. I will provide a method of reproducing the case later and add a new jira issue. > 2. What would be the repercussions if we have these out-of-order ops? Any > corruption? I think, kudu currently has ignored the problem or think it not a problem. I don't think that is not correct, we should talk about that. > 3. Any performance impact due to this serialisation? Yes, I will take a little cost about performance, but I think it's will be no confluence about that. The real performance impact, we should show the experiment metrics. http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@7 PS1, Line 7: keep the order of appling ops the same as raft wal > nit: Keep the order of applying the ops same as raft WAL. Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@10 PS1, Line 10: be the same > nit: be same Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@10 PS1, Line 10: Apply > nit: Applying Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@9 PS1, Line 9: is an total order : in server-side > nit: is in total order on server side Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@11 PS1, Line 11: But now apply ops by 'apply_pool->Submit' that's a concurrent threadpool, > nit: Currently, applying ops using "apply_pool_->Submit" (i.e. concurrent t Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@12 PS1, Line 12: some ops out > nit: some ops to go out Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@29 PS1, Line 29: would valid > nit: would become valid Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@29 PS1, Line 29: > nit: extra whitespace Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@29 PS1, Line 29: not our expected > nit: not expected Done http://gerrit.cloudera.org:8080/#/c/19463/1//COMMIT_MSG@30 PS1, Line 30: would > nit: could Done -- To view, visit http://gerrit.cloudera.org:8080/19463 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Id040122ab9473d1a8b03a2a1e751454a32b89d87 Gerrit-Change-Number: 19463 Gerrit-PatchSet: 1 Gerrit-Owner: Yuqi Du <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Ashwani Raina <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Yingchun Lai <[email protected]> Gerrit-Reviewer: Yuqi Du <[email protected]> Gerrit-Comment-Date: Fri, 03 Feb 2023 12:05:46 +0000 Gerrit-HasComments: Yes
