| @clarkb:matrix.org | I suspect either https://github.com/PyCQA/pycodestyle/commit/f04e2a1f5d37f8668519f123b97bf38165994b72 or https://github.com/PyCQA/pycodestyle/commit/6fddf7399d70627c46d1cc82bb3c02da2d708ec4 are to blame for the pycodestyle thing. That said maybe the lexer isn't consuming the \ properly instead due to some other change resulting in `next_char in WHITESPACE` checks breaking | 05:06 |
|---|---|---|
| @clarkb:matrix.org | I don't think this is intentional fwiw | 05:07 |
| @clarkb:matrix.org | that said the easiest thing is probably to add a space between `,` and `/` otherwise we're waiting or pinning. | 05:10 |
| @jjbeckman:matrix.org | Hi folks, would appreciate some advice. I am running Zuul on a Kubernetes cluster, and I find the following error happening seemingly randomly in various parts of the pipeline. | 08:15 |
| ``` | ||
| 2023-07-31 08:04:12.621503 | TASK [Read target {redacted} metadata file] | ||
| 2023-07-31 08:06:16.690816 | debian-bullseye | ERROR | ||
| 2023-07-31 08:06:16.691120 | debian-bullseye | { | ||
| 2023-07-31 08:06:16.691162 | debian-bullseye | "msg": "failed to transfer file /var/lib/zuul/builds/3f9565e3e3c1474c8103df5247a64576/work/tmp/ansible-local-2rastc7sj/tmp8v1_trs9 to '/~zuul/.ansible/tmp/ansible-tmp-1690790653.9951165-654-214536513211330/AnsiballZ_command.py':\nb''\nb'error: error sending request: Post \"https://10.0.0.1:443/api/v1/namespaces/main-0000000250/pods/debian-bullseye/exec?command=%2Fbin%2Fsh&command=-c&command=dd+of%3D%27%2F~zuul%2F.ansible%2Ftmp%2Fansible-tmp-1690790653.9951165-654-214536513211330%2FAnsiballZ_command.py%27+bs%3D65536+%26%26+sleep+0&container=debian-bullseye&stderr=true&stdin=true&stdout=true\": unexpected EOF\\n'" | ||
| 2023-07-31 08:06:16.691191 | debian-bullseye | } | ||
| ``` | ||
| Although probably unrelated, this particular error occured in this job, which almost always just executes fine. | ||
| ``` | ||
| - name: "Read target {redacted} metadata file" | ||
| ansible.builtin.command: "cat /{{ zuul.project.src_dir }}/target-{redacted}.yaml" | ||
| register: target_{redacted} | ||
| changed_when: false | ||
| ``` | ||
| Note, this only happens once in a dozen executions. | ||
| Would there be anyone here familiar with this error, and know how to solve the issue? | ||
| @clarkb:matrix.org | The AnsibleZ you see there refers to the file that Ansible copies to the remote host to execute there | 08:42 |
| @clarkb:matrix.org | It looks like you are running your test nodes as k8s resources as well since that isn't an ssh based copy but instead an http post | 08:42 |
| @clarkb:matrix.org | I would look into the kubectl copy method. https://stackoverflow.com/questions/67624630/unable-to-copy-data-from-pod-to-local-using-kubectl-cp-command stuff like this may be helpful | 08:43 |
| @jjbeckman:matrix.org | Thank you for the advice Clark. | 08:55 |
| > It looks like you are running your test nodes as k8s resources as well since that isn't an ssh based copy but instead an http post | ||
| Yes, my test nodes are Kubernetes pods. | ||
| > I would look into the kubectl copy method. | ||
| I'm unsure what you mean by "look into the kubectl copy method". I've copied all source code to the node pod using `kubectl cp`(Actually, `oc rsync` to be exact), but randomly get this "failed to transfer file" in a job. Could you please explain what you mean when you say to "look into the kubectl copy method", in this context? | ||
| @clarkb:matrix.org | > <@jjbeckman:matrix.org> Thank you for the advice Clark. | 09:46 |
| > | ||
| > > It looks like you are running your test nodes as k8s resources as well since that isn't an ssh based copy but instead an http post | ||
| > | ||
| > Yes, my test nodes are Kubernetes pods. | ||
| > | ||
| > > I would look into the kubectl copy method. | ||
| > | ||
| > I'm unsure what you mean by "look into the kubectl copy method". I've copied all source code to the node pod using `kubectl cp`(Actually, `oc rsync` to be exact), but randomly get this "failed to transfer file" in a job. Could you please explain what you mean when you say to "look into the kubectl copy method", in this context? | ||
| > | ||
| The Ansible kubernetes connection method is using the same protocol to copy the code it will execute on the test node. It isn't using ssh/rsync. I'm suggesting you look into the Ansible use of this system to debug it | ||
| @wils93:matrix.org | Hi everyone 👏 | 11:28 |
| I'm new here, currently checking this awesome tool. | ||
| I have a question in mind but not sure that I find the information under the docs, the question is: is it possible to define a custom merge command other than the normal git workflow? | ||
| The use case is that I have a repo that contains some configs files that aren't mergeable by default but needs some custom scripts in order to get all things in place | ||
| @jim:acmegating.com | wils93: hi and welcome! that's not possible, and unlikely to be supported since zuul needs to control all aspects of the git repo in order to implement its speculative gating functions. however, you can manipulate the repo at the start of the job, after the git merge. | 13:17 |
| @wils93:matrix.org | > <@jim:acmegating.com> wils93: hi and welcome! that's not possible, and unlikely to be supported since zuul needs to control all aspects of the git repo in order to implement its speculative gating functions. however, you can manipulate the repo at the start of the job, after the git merge. | 13:36 |
| okaay, so by that if the merge fails I can't workaround this to the manipulation part. correct? | ||
| @jim:acmegating.com | wils93: correct; zuul wants to be in control of merges, so if the change isn't git-mergeable then it knows it's already not going to be able to merge. | 13:40 |
| @jim:acmegating.com | * wils93: correct; zuul wants to be in control of merges, so if the change isn't git-mergeable then it knows it's already not going to be able to be merged by the code review system. | 13:41 |
| @bridgefan:matrix.org | Hi! I also have a question semi-related to merges. I noticed an interesting observation for our Zuul instance. We had a scenario where we were getting "This change depends on a change that failed to merge" errors on a repo-branch combination that our triggers don't cover. The repo has triggers for jobs but those triggers have branch filters that don't include the branch that we were getting these merge errors on. We understand why the merge error occurs (a scenario where a dependent change was in the process of merging when this commit tried to run) but weren't expecting Zuul to look at this branch at all. Do merge errors happen before the branch filters are checked? Is this expected? | 14:51 |
| @jim:acmegating.com | bridgefan: yes, and that's partially to enable the ability to speculatively change configuration (zuul has to merge the change to decide if it's a change that adds a new job) | 14:54 |
| @bridgefan:matrix.org | corvus: Thanks a lot for confirming our theory. | 14:56 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 890166: Add gcloud pubsub support to Gerrit driver https://review.opendev.org/c/zuul/zuul/+/890166 | 22:06 | |
| @vlotorev:matrix.org | Hi, it seems zuul doesn't check semaphore existence when job references semaphore. | 22:35 |
| Here is an example in sanbox repo https://review.opendev.org/c/opendev/sandbox/+/890119/6/zuul.yaml. | ||
| This change specifies job to check pipeline and defines `semaphores: some-unexisting-semaphore`, this change is not reported by zuul as configuration error (actually it can be submitted without errors on zuul side). | ||
| @vlotorev:matrix.org | * Hi, it seems zuul doesn't check semaphore existence when job references semaphore. | 22:38 |
| Here is an example in sanbox repo https://review.opendev.org/c/opendev/sandbox/+/890119/6/zuul.yaml. | ||
| This change specifies job to check pipeline and job references to `semaphores: some-unexisting-semaphore`, this change is not reported by zuul as configuration error (actually it can be submitted without errors on zuul side). | ||
| @jim:acmegating.com | vlotorev: correct, semaphores referenced by name are implicitly created if they don't exist | 22:48 |
| @vlotorev:matrix.org | hm, once merged such implicit semaphore is not shown on web UI. Also implicit semaphores not specified in doc. Is it okay to update the doc? | 22:52 |
| @jim:acmegating.com | vlotorev: yes, i think a doc update would be a good idea, thanks | 22:53 |
| @jim:acmegating.com | i would like to remove the implicit semaphore creation in the future, but it's going to be a potentially big and disruptive upgrade, so needs to be handled carefully. i think we just about have what we need to get started on that (the new configuration warnings should help). | 22:57 |
| @jim:acmegating.com | vlotorev: to be clear, for the docs update -- the implicit semaphore is created with max=1 | 22:58 |
| @jim:acmegating.com | (the default) | 22:59 |
| -@gerrit:opendev.org- lotorev vitaly proposed: [zuul/zuul] 890065: doc: Specify implicitly created semaphores https://review.opendev.org/c/zuul/zuul/+/890065 | 23:06 | |
| -@gerrit:opendev.org- lotorev vitaly proposed: [zuul/zuul] 890065: doc: Specify implicitly created semaphores https://review.opendev.org/c/zuul/zuul/+/890065 | 23:07 | |
| -@gerrit:opendev.org- lotorev vitaly proposed: [zuul/zuul] 890065: doc: Specify implicitly created semaphores https://review.opendev.org/c/zuul/zuul/+/890065 | 23:08 | |
| -@gerrit:opendev.org- lotorev vitaly proposed: [zuul/zuul] 890065: doc: Specify implicitly created semaphores https://review.opendev.org/c/zuul/zuul/+/890065 | 23:09 | |
| @vlotorev:matrix.org | corvus: ^ is a change updating the doc. | 23:12 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!