Hi Quanlong, I added a parameter to the jenkins job. This test run looks like it works ok: https://jenkins.impala.io/job/cherrypick-2.x-and-test/681/console
Please have a look and let me know if you require any further changes. Cheers, Lars On Sat, Jun 15, 2019 at 11:36 PM Quanlong Huang <[email protected]> wrote: > It happens again that a clean pick introduces build failure: > > - Jenkins run: > https://jenkins.impala.io/job/cherrypick-2.x-and-test/678/ > - Commit required some fix: > https://github.com/apache/impala/commit/c333b5526 > > Now the HEAD of 2.x branch is e6c1eb85eb31264eaf5ed92782bf181225ce9581. > There are 20 commits that can be cleanly applied to the 2.x branch before > this commit (c333b5526). It'd be better that the Jenkins job can pick them > first. Then we just need to submit one patch for review, requiring pretty > less approvals than the last time. So we can move forward faster. > > I submitted a patch for giving a terminal commit hash to > bin/compare_branches.py: https://gerrit.cloudera.org/c/13660 > If this got approved, hope someone have the permission can do me a favor to > add an optional parameter for this terminal commit hash in > the cherrypick-2.x-and-test Jenkins job. > > Thanks, > Quanlong > > > On Wed, Mar 20, 2019 at 10:12 AM Quanlong Huang <[email protected]> > wrote: > > > Sure. I just uploaded and published a batch of patches: > > > > remote: New Changes: > > > > remote: http://gerrit.cloudera.org:8080/12798 IMPALA-6642 (Part 2): > > clean up start-impala-cluster.py [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12799 IMPALA-6988: Implement > > ALTER TABLE/VIEW SET OWNER [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12800 IMPALA-6223: Gracefully > > handle malformed 'with' queries in impala-shell [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12801 IMPALA-6031: Fix > executor > > node count in distributed plans [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12802 IMPALA-3956: [DOCS] > Escape > > variables with '\' in impala-shell [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12803 IMPALA-6918: [DOCS] > > COMMENT ON COLUMN privileges [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12804 [DOCS] Removed an unused > > keydef [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12805 [DOCS] A typo fixed in > > impala_shell_running_commands [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12806 Fix zsh issue in > > set-pythonpath.sh [DRAFT] > > > > remote: http://gerrit.cloudera.org:8080/12807 IMPALA-7254: > Inconsistent > > decimal behavior for IN/BETWEEN predicate [DRAFT] > > > > Note that I've ignored c01efd096 in master branch since it's reverted > > later (df78eaec0). In all the above patches, just the last one > > (IMPALA-7254) needs some changes to fix test failures. I've added what I > > changed in the review comments. Hope anyone has time can help to review > > them. > > > > Thanks, > > Quanlong > > > > > > On Tue, Mar 19, 2019 at 4:22 AM Lars Volker <[email protected]> wrote: > > > >> Yes, you could have a local branch, cherry pick all the changes that you > >> know are working plus the last one that causes issues. Then you can send > >> the whole branch for review, which will open a new review for each of > the > >> changes and then submit them manually. > >> > >> Cheers, Lars > >> > >> On Fri, Mar 15, 2019 at 6:09 PM Quanlong Huang <[email protected] > > > >> wrote: > >> > >> > I thought all the submits to 2.x and master branch should be done in > >> > Gerrit. Do you mean I can create a new branch manually with the next > 10 > >> > patches, and then merge it *manually* to branch 2.x after > >> > parallel-all-tests run successfully on it? > >> > > >> > -- > >> > Quanlong > >> > > >> > On Sat, Mar 16, 2019 at 12:17 AM Lars Volker <[email protected]> wrote: > >> > > >> > > The Jenkins job uses compare_branches.py > >> > > <https://github.com/apache/impala/blob/2.x/bin/compare_branches.py> > >> to > >> > do > >> > > the cherry picking. That script takes source_branch and > target_branch > >> > > parameters, but they need to be valid keys into the ignored commits > >> file. > >> > > You'd probably want to amend that script to take a commit up until > >> which > >> > > you want to pick. Alternatively you could do the picking manually by > >> > > passing a list of the commits from your spreadsheet into "git > commit" > >> (it > >> > > can handle multiple commits at once). Then just run > >> parallel-all-tests on > >> > > it. > >> > > > >> > > Let me know if you need help with triggering any of this. > >> > > > >> > > Cheers, Lars > >> > > > >> > > On Fri, Mar 15, 2019 at 8:37 AM Quanlong Huang < > >> [email protected]> > >> > > wrote: > >> > > > >> > > > Hi all, > >> > > > > >> > > > Branch 2.x is now at 193a1b5. See more details: > >> > > > > >> > > > > >> > > > >> > > >> > https://docs.google.com/spreadsheets/d/12h1rTAPS1gm0vhlDGxeOXjnRD7rrOcoqzX4rjRRCyBg > >> > > > > >> > > > I got a trouble that the cherrypick-2.x-and-test job is able to > >> cleanly > >> > > > pick the next 15 patches but build fail due to a patch in the > middle > >> > > > (f16436628). It needs some changes to fix the test. But I can't > >> submit > >> > > the > >> > > > new patch until branch 2.x has picked the patch before it. > >> > > > > >> > > > Is it possible to add a parameter in job cherrypick-2.x-and-test, > to > >> > tell > >> > > > it that don't consider patches after a commit? So I can let > >> > > > cherrypick-2.x-and-test > >> > > > pick only the next 10 patches and submit a new patch for the fix > >> later. > >> > > > > >> > > > Thanks, > >> > > > Quanlong > >> > > > > >> > > > >> > > >> > > >
