| -@gerrit:opendev.org- Zuul merged on behalf of Dong Zhang: [zuul/zuul] 875633: Ignore fetch-ref-replicated gerrit event https://review.opendev.org/c/zuul/zuul/+/875633 | 10:49 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 875970: Fix race related to PR with changed base branch https://review.opendev.org/c/zuul/zuul/+/875970 | 11:23 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 875970: Fix race related to PR with changed base branch https://review.opendev.org/c/zuul/zuul/+/875970 | 11:25 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: | 12:14 | |
| - [zuul/zuul] 875574: Set layout state event ltime in delete-pipeline-state https://review.opendev.org/c/zuul/zuul/+/875574 | ||
| - [zuul/zuul] 875575: Avoid layout updates after delete-pipeline-state https://review.opendev.org/c/zuul/zuul/+/875575 | ||
| -@gerrit:opendev.org- daniel.pawlik https://matrix.to/#/@dpawlik:matrix.org proposed: [zuul/zuul-jobs] 876081: Provide deploy-microshift role https://review.opendev.org/c/zuul/zuul-jobs/+/876081 | 13:14 | |
| @flaper87:matrix.org | I added a custom logs url and the generated URL points to an `index.html` file in a private GCS bucket. I noticed that Zuul is adding a [trailing slash to the log_url, ignoring that it points to a specific file](https://opendev.org/zuul/zuul/src/branch/master/zuul/model.py#L3995-L3999) (Unless I read the wrong code, of course). | 13:32 |
|---|---|---|
| Is this done for a specific reason? I'm wondering if this part of the code should be removed entirely (the one that adds the slash) or just add a check for a file extension. I would understand if that has to be kept for backwards compatibility purposes but it should really be the job's responsibility to return the correct URL. | ||
| (this is all assuming I did things correctly, of course :D ) | ||
| @sean-k-mooney:matrix.org | hi o/ just an fyi you might see https://review.opendev.org/c/zuul/zuul-jobs/+/876081. its adding a new role to deploy microshift (an edge optimised lightweight version of openshift) | 13:35 |
| @sean-k-mooney:matrix.org | this is a port of a role that is being devepleped currenlty for software factory/rdo usage | 13:36 |
| @sean-k-mooney:matrix.org | but i alked about this a few month ago and suggested it woudl be better to upstream it to zuul jobs | 13:36 |
| @sean-k-mooney:matrix.org | form a zuul perspective it will provie a simpel way to test zuul with an openshift 4 like envionment | 13:37 |
| @sean-k-mooney:matrix.org | while its really just a hope at this point im hoping we could use microshfit to do some testing with the new installer redhat is working on to replace tripleo | 13:38 |
| @sean-k-mooney:matrix.org | it is currently in developer preview https://microshift.io/ and will evenutaly become part of OKD as well but the role might still be useful before then | 13:40 |
| @sean-k-mooney:matrix.org | the system requirements https://microshift.io/docs/getting-started/#system-requirements are one of the reasons its compleing for ci and test vs other openshift installation methods however as it has a narrow scope then openshift it also has less features | 13:42 |
| @sean-k-mooney:matrix.org | anyway i just wanted to provide some context for https://review.opendev.org/c/zuul/zuul-jobs/+/876081 | 13:42 |
| -@gerrit:opendev.org- Flavio Percoco proposed: [zuul/zuul] 876112: Don't append trailing slash if the URL points to a file https://review.opendev.org/c/zuul/zuul/+/876112 | 13:58 | |
| @flaper87:matrix.org | > <@gerrit:opendev.org> Flavio Percoco proposed: [zuul/zuul] 876112: Don't append trailing slash if the URL points to a file https://review.opendev.org/c/zuul/zuul/+/876112 | 13:59 |
| Just being proactive for the sake of discussion. Unsure if this is actually the desired behavior here | ||
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 876159: Don't discard all cat job results in case of error https://review.opendev.org/c/zuul/zuul/+/876159 | 14:32 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 876159: Don't discard all cat job results in case of error https://review.opendev.org/c/zuul/zuul/+/876159 | 14:35 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 876159: Don't discard all cat job results in case of error https://review.opendev.org/c/zuul/zuul/+/876159 | 14:36 | |
| -@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 876159: Don't discard all cat job results in case of error https://review.opendev.org/c/zuul/zuul/+/876159 | 14:37 | |
| @jim:acmegating.com | flaper87: log_url should never point to a file. | 14:40 |
| @fungicide:matrix.org | though i wonder if there are object storage platforms which don't support either default files or directory-like generated indices, i have a feeling gcs supports at least one of those like swift and s3 do | 14:43 |
| @jim:acmegating.com | fungi: sure, but that's not the right solution; that breaks all the web ui log and artifact handling | 14:46 |
| @vonschultz:matrix.org | > <@fungicide:matrix.org> though i wonder if there are object storage platforms which don't support either default files or directory-like generated indices, i have a feeling gcs supports at least one of those like swift and s3 do | 14:58 |
| For S3, we've ended up using nginx-s3-gateway with `PROVIDE_INDEX_PAGE=1`. That way, we don't need to have a public bucket, and the default file is handled by the gateway. I suspect there are other object storage platforms that would also need something similar to work well here. | ||
| @clarkb:matrix.org | Do private buckets break the uploaded indexes? I guess because you don't know what the root is maybe? | 15:12 |
| @flaper87:matrix.org | > <@jim:acmegating.com> flaper87: log_url should never point to a file. | 15:19 |
| can you elaborate? Not even the `index.html` ? I have a private bucket that I'm accessing through a proxy, which goes and reads files and sends output back. It's a bit strange that I can't point it to `index.html` directly | ||
| @flaper87:matrix.org | I'm definitely missing something :) | 15:19 |
| @flaper87:matrix.org | Especially because I don't see anything in the `logs` secion of the UI. Although the `View Log` link actually works | 15:19 |
| @flaper87:matrix.org | > <@vonschultz:matrix.org> For S3, we've ended up using nginx-s3-gateway with `PROVIDE_INDEX_PAGE=1`. That way, we don't need to have a public bucket, and the default file is handled by the gateway. I suspect there are other object storage platforms that would also need something similar to work well here. | 15:21 |
| I endeed up doing something similar, but this requires more configs on the nginx side. (basically a rewrite of the path, which is not great but works) | ||
| @jim:acmegating.com | flaper87: ignore the log link on the page. the real fix is that should be deleted. think of it as "debug log link" | 15:21 |
| @jim:acmegating.com | as i said, the log_url should point to a directory, and zuul-web will read the artifacts, including the manifest file from that location, and fetch everything else | 15:22 |
| @flaper87:matrix.org | mmh, I see. I understand why this was done. It's actually not happening, I'd guess because the gcs bucket is private and I'm accessing it through a gcs-proxy which doesn't list the artifacts in the dir | 15:25 |
| @jim:acmegating.com | nope that's not it at all | 15:25 |
| @jim:acmegating.com | well, that's half right | 15:26 |
| @jim:acmegating.com | zuul-web doesn't care about listing things | 15:26 |
| @jim:acmegating.com | zuul-web just needs to fetch the manifest file which is a zuul artifact from the log url | 15:27 |
| @flaper87:matrix.org | ah, I get it! thanks for the details. I see the manifest and it's correct. Now, the interesting thing is that the `logs` tab is empty, which I was expecting not to be. | 15:31 |
| @jim:acmegating.com | check the browser devtools for errors related to loading the manifest | 15:32 |
| @flaper87:matrix.org | alrighty, thanks :) | 15:33 |
| @flaper87:matrix.org | semi-related, the zuul jobs running on pods are basically only producing the following output: | 15:34 |
| ``` | ||
| 2023-03-02 15:08:48.180076 | [pod-fedora] Waiting on logger | ||
| 2023-03-02 15:08:48.424356 | [pod-fedora] Waiting on logger | ||
| 2023-03-02 15:08:48.777666 | [Zuul] Log Stream did not terminate | ||
| 2023-03-02 15:08:48.777979 | pod-fedora | ERROR | ||
| 2023-03-02 15:08:48.778169 | pod-fedora | { | ||
| 2023-03-02 15:08:48.778236 | pod-fedora | "delta": "0:00:11.880583", | ||
| 2023-03-02 15:08:48.778290 | pod-fedora | "end": "2023-03-02 15:06:48.552924", | ||
| 2023-03-02 15:08:48.778320 | pod-fedora | "msg": "non-zero return code", | ||
| 2023-03-02 15:08:48.778343 | pod-fedora | "rc": 1, | ||
| 2023-03-02 15:08:48.778366 | pod-fedora | "start": "2023-03-02 15:06:36.672341" | ||
| 2023-03-02 15:08:48.778387 | pod-fedora | } | ||
| 2023-03-02 15:08:48.781683 | | ||
| ``` | ||
| The waiting on logger message appears pretty much for every task. Unsure if this is a known/expected behavior when using `kubectl` as a connection | ||
| -@gerrit:opendev.org- daniel.pawlik https://matrix.to/#/@dpawlik:matrix.org proposed: [zuul/zuul-jobs] 876081: Provide deploy-microshift role https://review.opendev.org/c/zuul/zuul-jobs/+/876081 | 15:48 | |
| @vonschultz:matrix.org | > <@clarkb:matrix.org> Do private buckets break the uploaded indexes? I guess because you don't know what the root is maybe? | 15:52 |
| No, I think the uploaded indexes are fine. And I can do some normal nginx redirects to get S3 logs where I want them in the web interface. But unless you enable "Static website hosting" for the bucket it won't redirect a directory to the corresponding index document. So something more intelligent was needed, such as the nginx-s3-gateway, for the bucket to remain private. `PROVIDE_INDEX_PAGE=1` "returns the index page if there is one when requesting a directory. Cannot be enabled with `ALLOW_DIRECTORY_LIST`." (https://github.com/nginxinc/nginx-s3-gateway/blob/master/docs/getting_started.md), so it's not that the uploaded indexes are wrong in any way, just that we need to make sure they get used. | ||
| @clarkb:matrix.org | got it thanks for the clarification | 15:53 |
| @clarkb:matrix.org | sean-k-mooney: dpawlik it would probably be a good idea to include tests for that role with the new role so that we can see it working. That should all be self testing | 15:56 |
| @clarkb:matrix.org | But ya that could be useful to update nodepool openshift testing | 15:59 |
| @clarkb:matrix.org | would be another good place to run it to prove it is working as expected | 15:59 |
| @sean-k-mooney:matrix.org | it would be nice to test it with molecue | 16:03 |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 859939: Implement Gitea driver https://review.opendev.org/c/zuul/zuul/+/859939 | 16:21 | |
| -@gerrit:opendev.org- Artem Goncharov proposed: [zuul/zuul] 859940: Implement tests of the gitea driver https://review.opendev.org/c/zuul/zuul/+/859940 | 16:21 | |
| -@gerrit:opendev.org- Matthieu Huin https://matrix.to/#/@mhuin:matrix.org proposed: [zuul/zuul] 872552: Keycloak tutorial: update to keycloak 20, add CLI example https://review.opendev.org/c/zuul/zuul/+/872552 | 16:35 | |
| @fungicide:matrix.org | > <@jim:acmegating.com> fungi: sure, but that's not the right solution; that breaks all the web ui log and artifact handling | 16:47 |
| oh, right this is about the base url where the webui should find the manifest; the manifest *is the* file index as far as the webui is concerned | ||
| @flaper87:matrix.org | > <@fungicide:matrix.org> oh, right this is about the base url where the webui should find the manifest; the manifest *is the* file index as far as the webui is concerned | 17:04 |
| TIL :) thanks y'all | ||
| @flaper87:matrix.org | corvus: re the logs not showing up on the web UI, it seems like it may be a problem with the oauth proxy I've got in front of the GCS proxy. Basically, the web client is trying to access the files but they are behind oauth. | 17:07 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 875790: Add Gerrit pipeline trigger requirements https://review.opendev.org/c/zuul/zuul/+/875790 | 21:22 | |
| -@gerrit:opendev.org- Christian Mueller proposed: [zuul/nodepool] 865531: Provide the option to launch Amazon EC2 Spot instances https://review.opendev.org/c/zuul/nodepool/+/865531 | 22:01 | |
| -@gerrit:opendev.org- Christian Mueller proposed: [zuul/nodepool] 865531: Provide the option to launch Amazon EC2 Spot instances https://review.opendev.org/c/zuul/nodepool/+/865531 | 22:05 | |
| -@gerrit:opendev.org- Zuul merged on behalf of James E. Blair https://matrix.to/#/@jim:acmegating.com: [zuul/nodepool] 871199: Add API timing debug statements to openstack driver https://review.opendev.org/c/zuul/nodepool/+/871199 | 22:35 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!