| -@gerrit:opendev.org- Simon Westphahl proposed on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 831430: Dequeue items that can no longer merge https://review.opendev.org/c/zuul/zuul/+/831430 | 06:08 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 831460: Handle Github branch protection rule webhook events https://review.opendev.org/c/zuul/zuul/+/831460 | 07:25 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 830840: Make it possible to configure job retries with zuul_return https://review.opendev.org/c/zuul/zuul/+/830840 | 09:12 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 830840: Make it possible to configure job retries with zuul_return https://review.opendev.org/c/zuul/zuul/+/830840 | 09:13 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 832361: Support submitWholeTopic in Gerrit https://review.opendev.org/c/zuul/zuul/+/832361 | 09:15 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Dong Zhang: [zuul/zuul] 831925: Inject bundle id to inventroy zuul.items https://review.opendev.org/c/zuul/zuul/+/831925 | 09:28 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 830840: Make it possible to configure job retries with zuul_return https://review.opendev.org/c/zuul/zuul/+/830840 | 10:30 | |
| -@gerrit:opendev.org- Albin Vass marked as active: [zuul/zuul] 830840: Make it possible to configure job retries with zuul_return https://review.opendev.org/c/zuul/zuul/+/830840 | 10:30 | |
| -@gerrit:opendev.org- Simon Westphahl proposed on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 831873: Use JobData for Build result data https://review.opendev.org/c/zuul/zuul/+/831873 | 11:26 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 830840: Make it possible to configure job retries with zuul_return https://review.opendev.org/c/zuul/zuul/+/830840 | 11:54 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Dong Zhang: [zuul/zuul] 827788: Not swallow exception when cat job fails https://review.opendev.org/c/zuul/zuul/+/827788 | 12:19 | |
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/zuul] 834671: WIP: Handle reviews by anonymous github users https://review.opendev.org/c/zuul/zuul/+/834671 | 12:29 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/zuul] 832362: Report per-branch cyclic-dependency conflicts https://review.opendev.org/c/zuul/zuul/+/832362 | 12:38 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Dong Zhang: [zuul/zuul] 830628: Items in extra paths should be loaded in dependent changes https://review.opendev.org/c/zuul/zuul/+/830628 | 12:38 | |
| @q:fricklercloud.de | does anyone actually use zuul-triggered merges on github? I'm seeing so many weird issues that I start to doubt it | 13:17 |
|---|---|---|
| like look at https://github.com/osism/ansible-defaults/commit/a7b365ba45880bb858f1b72eacddfe26a6c204dd, the commit title is duplicated and the "<>" are stripped from the commit message | ||
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/zuul] 834687: Do not duplicate the commit title on github https://review.opendev.org/c/zuul/zuul/+/834687 | 13:28 | |
| -@gerrit:opendev.org- Dr. Jens Harbott proposed: [zuul/zuul] 834671: WIP: Handle reviews by anonymous github users https://review.opendev.org/c/zuul/zuul/+/834671 | 14:41 | |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 834693: Bugfix: fix extra-config-paths for subdirectories https://review.opendev.org/c/zuul/zuul/+/834693 | 14:43 | |
| @fungicide:matrix.org | > <@q:fricklercloud.de> does anyone actually use zuul-triggered merges on github? I'm seeing so many weird issues that I start to doubt it | 14:45 |
| > like look at https://github.com/osism/ansible-defaults/commit/a7b365ba45880bb858f1b72eacddfe26a6c204dd, the commit title is duplicated and the "<>" are stripped from the commit message | ||
| could that just be github's ui at fault? here's the actual commit in patch form as served from github: https://github.com/osism/ansible-defaults/commit/a7b365ba45880bb858f1b72eacddfe26a6c204dd.patch | ||
| @fungicide:matrix.org | also it looks like the dco signed-off-by and the reviewed-by footers got split into separate paragraphs instead of both being in the footer. i wonder if that has anything to do with the missing <> | 14:47 |
| -@gerrit:opendev.org- Albin Vass proposed: [zuul/zuul] 834693: Fix extra-config-paths for non root directories https://review.opendev.org/c/zuul/zuul/+/834693 | 14:47 | |
| @frickler:freitrix.de | that .patch link does show both issues like the UI, not sure what else you see there | 14:47 |
| @frickler:freitrix.de | the \n\n between the commit_message body and the added footers is intentionally added by zuul https://opendev.org/zuul/zuul/src/branch/master/zuul/driver/github/githubreporter.py#L331 | 14:49 |
| @fungicide:matrix.org | oh, zuul created that commit, not you? | 14:49 |
| @fungicide:matrix.org | did it rebase your commit when merging or something? | 14:49 |
| @fungicide:matrix.org | also i see what you mean now about the doubled sumbect, it's in the subject header but then also included in the first line of the body | 14:50 |
| @clarkb:matrix.org | I thought zuul just clicks the merge button | 14:50 |
| @frickler:freitrix.de | no, that's what zuul does when merging. the original commit submitted for the PR is https://github.com/osism/ansible-defaults/pull/65/commits/19e69f6fe73d2531abba97422db7273cfeaaa0f8 | 14:50 |
| @fungicide:matrix.org | * also i see what you mean now about the doubled subbect, it's in the subject header but then also included in the first line of the body | 14:51 |
| @frickler:freitrix.de | zuul amends the commit message, see the function around the line I linked and the two patches I submitted earlier | 14:52 |
| @fungicide:matrix.org | yeah, found the reviewed-by addition in this function: https://opendev.org/zuul/zuul/src/branch/master/zuul/driver/github/githubreporter.py#L317-L334 | 14:54 |
| @fungicide:matrix.org | oh, you already linked to it as well | 14:55 |
| @frickler:freitrix.de | btw. I also didn't manage to run the unit tests locally. and 60-90 mins for that job is also ... wow https://zuul.opendev.org/t/zuul/builds?job_name=zuul-tox-py38&project=zuul/zuul | 14:57 |
| @fungicide:matrix.org | looks like it would need some sort of heuristic to be able to discern between a commit message with an existing footer and one without, since in the without case it would need to add a blank line before appending its reviewed-by lines, while in the with case it should not add a blank line first | 14:57 |
| @frickler:freitrix.de | > <@fungicide:matrix.org> looks like it would need some sort of heuristic to be able to discern between a commit message with an existing footer and one without, since in the without case it would need to add a blank line before appending its reviewed-by lines, while in the with case it should not add a blank line first | 14:59 |
| yes, because of that complication I didn't tackle that part so far | ||
| @fungicide:matrix.org | did you manage to spot why change.body_text is getting <> filtered out? | 14:59 |
| @frickler:freitrix.de | not yet. I did verify that the body does contain the <> in the github API response, but nothing more | 15:00 |
| @fungicide:matrix.org | suspicious we might be running it through some sort of html sanitizer (maybe inside requests itself) | 15:01 |
| @frickler:freitrix.de | curl -H "Accept: application/vnd.github.v3+json" https://api.github.com/repos/osism/ansible-defaults/pulls/65 | jq '.body' | 15:01 |
| @clarkb:matrix.org | > <@frickler:freitrix.de> btw. I also didn't manage to run the unit tests locally. and 60-90 mins for that job is also ... wow https://zuul.opendev.org/t/zuul/builds?job_name=zuul-tox-py38&project=zuul/zuul | 15:07 |
| Zuul's unittest framework is more like unittests with a lot of functional tests along side it. YOu need a zookeeper running and mysql and all the associated ssl certs. There are scripts in tools/ to help bootstrap this. But a lot of the time is due to the more functional nature of the testing | ||
| @fungicide:matrix.org | for not splitting the footer, it could probably re.match on something like ``'^[^ ]:'`` for the last line to decide if there's a preexisting footer | 15:08 |
| @fungicide:matrix.org | * for not splitting the footer, it could probably re.match on something like `'^[^ ]*:'` for the last line to decide if there's a preexisting footer | 15:09 |
| @frickler:freitrix.de | > <@clarkb:matrix.org> Zuul's unittest framework is more like unittests with a lot of functional tests along side it. YOu need a zookeeper running and mysql and all the associated ssl certs. There are scripts in tools/ to help bootstrap this. But a lot of the time is due to the more functional nature of the testing | 15:10 |
| I tried those and none of them worked. the tox-docker things seem helplessly outdated and with the docker-compose setup I got the same connect timeout issues that were mentioned the other day | ||
| @clarkb:matrix.org | Its been a while since I ran it but tools/test-setup-docker.sh is what worked for me in the past | 15:12 |
| @clarkb:matrix.org | I wonder if it is a zookeeper version thing since that docker-compose file doesn't pin zookeeper | 15:13 |
| @fungicide:matrix.org | looks like the github connection is getting the pr body_text from github3.pulls not directly from the github api, so filtering may be happening in there | 15:14 |
| @fungicide:matrix.org | i don't see anywhere that zuul is mangling the body_text on its own | 15:15 |
| @fungicide:matrix.org | digging around in github3.py source now to see if i can find it | 15:19 |
| @fungicide:matrix.org | no smoking gun yet, but the PullReview class has several related attributes: ``body`` is the pr body in markdown format, ``body_html`` is in html formatting, and ``body_text`` is "The plain-text formatted body of this review." (which given its proximity to the other two, suggests it may be defanged/sanitized in order to make it more "plain") | 15:25 |
| @fungicide:matrix.org | * no smoking gun yet, but the PullRequest class has several related attributes: `body` is the pr body in markdown format, `body_html` is in html formatting, and `body_text` is "The plain-text formatted body of this review." (which given its proximity to the other two, suggests it may be defanged/sanitized in order to make it more "plain") | 15:29 |
| @fungicide:matrix.org | the github api seems to just return "body" according to their api docs, so i'm still trying to figure out how github3 turns the body into body_text | 15:34 |
| @fungicide:matrix.org | bingo! https://docs.github.com/en/rest/overview/media-types#text | 15:50 |
| @fungicide:matrix.org | frickler: ``curl -H "Accept: application/vnd.github.v3.text+json" https://api.github.com/repos/osism/ansible-defaults/pulls/65 | jq '.body_text'`` | 15:53 |
| @fungicide:matrix.org | it's github's "text" conversion stripping the <> | 15:53 |
| @fungicide:matrix.org | apparently not all printable ascii is "text" in github's world | 15:55 |
| @fungicide:matrix.org | yeah, github3 sets ``Accept: application/vnd.github.v3.full+json`` when it makes requests, and that is how it gets all three of body, body_text and body_html | 15:58 |
| @frickler:freitrix.de | fungi: nice digging. we certainly don't want to use html, so the question is whether we can simply use body instead of body_text or whether we need some special treatment for that | 16:03 |
| @fungicide:matrix.org | the docs claim "body" is in markdown format | 16:04 |
| @fungicide:matrix.org | but what they mean by that is anybody's guess | 16:04 |
| @jim:acmegating.com | the pull request body | 16:04 |
| @frickler:freitrix.de | well I assume you can add markdown in the UI when editing the message for the PR | 16:04 |
| @fungicide:matrix.org | maybe that's their way of saying they don't "interpret" the markdown for you if you use "body" | 16:04 |
| @jim:acmegating.com | yeah, it's not the same thing as a commit message | 16:04 |
| @frickler:freitrix.de | well if the PR consists of a single commit, the initial body is copied from that commit's commit message | 16:05 |
| @frickler:freitrix.de | and the thing is called "commit_message" again when submitting the pr_merge request | 16:06 |
| @frickler:freitrix.de | let me put some markup into a PR and see how it looks in the API | 16:09 |
| @frickler:freitrix.de | it has \r\n instead of \n but otherwise I think this looks usable https://api.github.com/repos/osfrickler/testbed/pulls/5 | 16:18 |
| @frickler:freitrix.de | seems we currently also would lose the "- " prefixes for any line, which will be at least as confusing as losing the "<>" | 16:18 |
| @blaisep-sureify:matrix.org | > <@avass:vassast.org> yeah apparently that works, thanks! | 16:20 |
| Since I'm do neg the tutorial, I made this change to docs/source/examples/docker-compose.yaml: | ||
| ``` | ||
| gerritconfig: | ||
| image: docker.io/zuul/zuul-executor | ||
| environment: | ||
| - http_proxy | ||
| - https_proxy | ||
| - no_proxy=${no_proxy},gerrit | ||
| depends_on: | ||
| - gerrit | ||
| volumes: | ||
| - "sshkey:/var/ssh:z" | ||
| - "nodessh:/var/node:z" | ||
| - "./playbooks/:/var/playbooks/:z" | ||
| - "certs:/var/certs:z" | ||
| - "../../../tools/:/var/zuul-tools/:z" | ||
| # NOTE(pabelanger): Be sure to update this line each time we change the | ||
| # default version of ansible for Zuul. | ||
| command: "/usr/local/lib/zuul/ansible/2.9/bin/ansible-playbook /var/playbooks/setup.yaml \ | ||
| && sh -c 'zkServer.sh start-foreground'" | ||
| networks: | ||
| ``` | ||
| @blaisep-sureify:matrix.org | No sucess | 16:20 |
| @frickler:freitrix.de | I think the doc https://zuul-ci.org/docs/zuul/latest/config/project.html#value-project.merge-mode.squash-merge is wrong or at least confusing. "Squash merges each change onto the branch." sounds to me like every commit will end up in the branch individually, instead all commits are really squashed into a single commit that is added to the branch | 17:52 |
| @frickler:freitrix.de | also why is there no merge-mode that maps to github's "rebase"? or is that what "merge-resolve" really means? | 17:53 |
| @frickler:freitrix.de | when testing merging via github UI, it seems it does not use the PR body at all. instead one gets a template filled out that one can then edit, how it gets filled depends on the merge mode again | 17:54 |
| @fungicide:matrix.org | yeah, so to summarize, the underlying design problem is that the commit message in a single-commit pr is making a lossy round-trip through github's pull request description and then stuffed back into the final commit message | 18:32 |
| @fungicide:matrix.org | this seems more like a github design short-coming, but maybe we can find reasonable workarounds | 18:33 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 18:33 | |
| - [zuul/zuul] 831430: Dequeue items that can no longer merge https://review.opendev.org/c/zuul/zuul/+/831430 | ||
| - [zuul/zuul] 831873: Use JobData for Build result data https://review.opendev.org/c/zuul/zuul/+/831873 | ||
| @vlotorev:matrix.org | Hi, I filed security issue to https://storyboard.openstack.org/#!/story/2009938, have zuul maint seen this? | 19:21 |
| @jim:acmegating.com | vlotorev: we have now, thanks! | 19:24 |
| @andrebeltrami:matrix.org | Hi all, we got stucked in our CI with the following task https://github.com/openstack/devstack/blob/master/roles/capture-system-logs/tasks/main.yaml. Anyone already have this issue, or have an idea what's going on. The task started but never finish. | 19:41 |
| @clarkb:matrix.org | > <@andrebeltrami:matrix.org> Hi all, we got stucked in our CI with the following task https://github.com/openstack/devstack/blob/master/roles/capture-system-logs/tasks/main.yaml. Anyone already have this issue, or have an idea what's going on. The task started but never finish. | 19:53 |
| You might try updating the shell task there to run `set -x` which will enable tracing to hopefully identify where it is stuck | ||
| @blaisep-sureify:matrix.org | > <@avass:vassast.org> yeah apparently that works, thanks! | 19:54 |
| * ...ok.. I will try that too, because I get this error: | ||
| @andrebeltrami:matrix.org | I will try it Clark , thank you... | 19:59 |
| -@gerrit:opendev.org- Guillaume Chauvel proposed: | 21:22 | |
| - [zuul/zuul] 732068: tutorial: Add "Use zuul jobs" https://review.opendev.org/c/zuul/zuul/+/732068 | ||
| - [zuul/zuul] 732069: tutorial: Add "gate pipeline" https://review.opendev.org/c/zuul/zuul/+/732069 | ||
| - [zuul/zuul] 732070: tutorial: Add "job secrets" https://review.opendev.org/c/zuul/zuul/+/732070 | ||
| -@gerrit:opendev.org- Zuul merged on behalf of Tristan Cacqueray: [zuul/zuul] 834570: Update axios to fix CVE-2022-0536 https://review.opendev.org/c/zuul/zuul/+/834570 | 22:46 | |
| -@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 832655: Support non-top-level dirs as extra config path https://review.opendev.org/c/zuul/zuul/+/832655 | 23:25 | |
| @jim:acmegating.com | it looks like the ibm cloud sdk module used by the ibmvpc driver just became incompatible with python 3.6: https://zuul.opendev.org/t/zuul/build/0c95b5b7151d4c8eb638df68f4e44a84 | 23:48 |
| ```AttributeError: module 'ssl' has no attribute 'TLSVersion'``` | ||
| @jim:acmegating.com | should we: | 23:50 |
| # pin to an older version | ||
| # add a python version skip to the unit test and update the docs saying it's not compatible with 3.6 | ||
| # or drop 3.6 support from zuul/nodepool (it's eol)? | ||
| @jim:acmegating.com | * should we: | 23:50 |
| * pin to an older version | ||
| * add a python version skip to the unit test and update the docs saying it's not compatible with 3.6 | ||
| * or drop 3.6 support from zuul/nodepool (it's eol)? | ||
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!