Philip Durbin 🚀 (Aug 10 2023 at 18:37): Philip Durbin 🚀 (Aug 10 2023 at 18:37):Let's discuss and troubleshoot #9769 which is about some strangeness we'll seeing with images tagged as "stable". They are being updated at unexpected times.
Philip Durbin 🚀 (Aug 10 2023 at 18:38):Previously I reported this at #containers > Exception while loading the app but I think we deserve a fresh topic. :big_smile:
Philip Durbin 🚀 (Aug 10 2023 at 19:18):@Oliver Bertuch is https://github.com/IQSS/dataverse/actions/runs/5813835717/job/15762398940 a potential smoking gun simply due to timing?
I'm comparing digests and I don't find a match between that run and what I'm seeing on Docker Hub.
Philip Durbin 🚀 (Aug 10 2023 at 19:18):I'm also not finding a match between what I see for docker images gdcc/dataverse --digests locally and digests I see on Docker Hub, though. So maybe I'm doing something wrong. :sweat_smile:
Philip Durbin 🚀 (Aug 10 2023 at 19:20):I was hoping this would be simple, that I'd be able to track a digest from the GitHub Action run to Docker Hub (both activity and tags) to my local machine. But that's not what I'm seeing.
Philip Durbin 🚀 (Aug 10 2023 at 19:21):The only digest that's lining up so far is "pushed the tag unstable with the digest sha256:8dcef6d4166cc to the repository gdcc/dataverse" from https://hub.docker.com/repository/docker/gdcc/dataverse/activity
It lines up with the digest I see locally:
$ docker images gdcc/dataverse --digests
REPOSITORY TAG DIGEST IMAGE ID CREATED SIZE
gdcc/dataverse unstable sha256:8dcef6d4166cc5b83aa94fb7c886e190fc70a1894cf489e52feb6b006371a673 1e19f9de87db 22 hours ago 739MB
Philip Durbin 🚀 (Aug 10 2023 at 19:22):But I don't see that digest at https://hub.docker.com/repository/docker/gdcc/dataverse/tags
Philip Durbin 🚀 (Aug 10 2023 at 19:22):Or at https://github.com/IQSS/dataverse/actions/runs/5813835717/job/15762398940
Philip Durbin 🚀 (Aug 11 2023 at 09:36):Over at #9769 we're looking at - if: ${{ github.event_name != 'pull_request' }}
Oliver Bertuch (Aug 11 2023 at 10:20):image.png
@Philip Durbin WDYT?
Oliver Bertuch (Aug 11 2023 at 10:24):Built with GraphViz, source in Sphinx. The layout is not perfect, but it's hard to arrange things
Philip Durbin 🚀 (Aug 11 2023 at 10:29):Wow. Brain hurting a bit. I switched to decaf but I might need to switch back for this. Overall, looks great!
Oliver Bertuch (Aug 11 2023 at 10:29):Well we are doing complicated things in there...
Oliver Bertuch (Aug 11 2023 at 10:30):If you want to mess around and edit, here's a link to an editor with the graph
Oliver Bertuch (Aug 11 2023 at 10:33):Maybe it's easier to read this way, but it takes up way more space.
image.png
Philip Durbin 🚀 (Aug 11 2023 at 10:43):Yes, this is better. I'll probably print it out.
Philip Durbin 🚀 (Aug 11 2023 at 10:43):We can probably enumerate the common cases
Philip Durbin 🚀 (Aug 11 2023 at 10:44): Philip Durbin 🚀 (Aug 11 2023 at 10:44): Philip Durbin 🚀 (Aug 11 2023 at 10:44): Philip Durbin 🚀 (Aug 11 2023 at 10:44):etc.
Philip Durbin 🚀 (Aug 11 2023 at 10:44):And trace them in the diagram. Make sure it makes sense.
Oliver Bertuch (Aug 11 2023 at 10:57):Technically, this is only the app push workflow. It's not depicting the others
Oliver Bertuch (Aug 11 2023 at 10:57):So it's not showing how the Maven Unit tests might trigger it
Oliver Bertuch (Aug 11 2023 at 10:57):It's not showing the PR comment workflow, that works slightly different
Philip Durbin 🚀 (Aug 11 2023 at 12:25):Is this showing before/current behavior? Is the bug we're fixing in the diagram? Or is this diagram aspirational and bug free?
Oliver Bertuch (Aug 11 2023 at 12:43):I did a fix locally in the push workflow and created a diagram to explain the flow's logic
Oliver Bertuch (Aug 11 2023 at 12:43):So it will describe the "current" behaviour as it will be after merging the PR
Oliver Bertuch (Aug 11 2023 at 12:44):
Oliver Bertuch (Aug 11 2023 at 12:44):Here's an updated diagram @Philip Durbin, hopefully easier to follow
Oliver Bertuch (Aug 11 2023 at 13:09):I pushed what I have so far: https://dataverse-guide--9772.org.readthedocs.build/en/9772/container/auto-usage.html
Philip Durbin 🚀 (Aug 11 2023 at 13:09):Am I forgetting any CI triggers?
Oliver Bertuch (Aug 11 2023 at 13:12):Depends on your definition of trigger. Some of these are technically the same
Philip Durbin 🚀 (Aug 11 2023 at 13:17):Let's call them actions that can lead to a trigger, then. :big_smile:
Philip Durbin 🚀 (Aug 11 2023 at 13:18):Does creating a release trigger a push to Docker Hub? What about a draft release?
Oliver Bertuch (Aug 11 2023 at 13:19):No. Releases are a very different kind of activitiy which we do not have any workflows for at the moment.
Philip Durbin 🚀 (Aug 11 2023 at 13:19):ok
Oliver Bertuch (Aug 11 2023 at 13:19):We only ever react to: push, pull_request, schedule and comments
Philip Durbin 🚀 (Aug 11 2023 at 13:20):I'm having trouble understanding "trigger was no push or on branches develop or master".
Philip Durbin 🚀 (Aug 11 2023 at 13:20):Should we do a quick call?
Oliver Bertuch (Aug 11 2023 at 13:20):Sure
Oliver Bertuch (Aug 11 2023 at 13:20):Plz help me rephrase that
Oliver Bertuch (Aug 11 2023 at 13:20):https://fz-juelich-de.zoom.us/j/62105154469?pwd=dSsxKzkzbkZVa01MOTdXV2l4UXh4UT09
Philip Durbin 🚀 (Aug 11 2023 at 13:20):https://harvard.zoom.us/my/pdurbin?pwd=em1WNUZGbnY2YjhxNEdSbjJJMXNSUT09
Oliver Bertuch (Aug 11 2023 at 13:21)::see_no_evil:
Philip Durbin 🚀 (Aug 11 2023 at 14:30):Updated version:
Oliver Bertuch (Aug 11 2023 at 14:37):I added some more notes from what we spoke about to https://github.com/IQSS/dataverse/pull/9772
Philip Durbin 🚀 (Aug 11 2023 at 14:39):Cool. I'll probably add a review. Nice diagram.
Oliver Bertuch (Aug 14 2023 at 10:46):Created https://github.com/IQSS/dataverse/pull/9777 as requested, containing just the fix
Oliver Bertuch (Aug 14 2023 at 10:46):https://github.com/IQSS/dataverse/pull/9776 contains the docs (and is a draft)
Philip Durbin 🚀 (Aug 14 2023 at 10:51):Yes. Perfect. I assigned the PR to myself. When I move things to "ready to QA" I don't get a chance to talk about them at standup (because we skip that column) so I left it in "in review" (which we don't skip) so I can. Thanks, @Oliver Bertuch !
Philip Durbin 🚀 (Aug 14 2023 at 17:41):#9777 was just merged!
Philip Durbin 🚀 (Aug 14 2023 at 17:42):Did it fix the bug?
Philip Durbin 🚀 (Aug 14 2023 at 19:30):I just pushed a branch under IQSS: https://github.com/IQSS/dataverse/tree/9260-solr9
Oliver Bertuch (Aug 14 2023 at 19:53):So that is an upstream branch. I see https://github.com/IQSS/dataverse/actions/runs/5859827266/job/15886911953 that publishing an image was skipped, which is what we wanted.
Oliver Bertuch (Aug 14 2023 at 19:54):@Philip Durbin did you open a PR yet? It _should_ push a preview image then...
Philip Durbin 🚀 (Aug 14 2023 at 19:55):Nope and that may not happen today for this branch. Still hacking.
Oliver Bertuch (Aug 14 2023 at 20:02):Will try with the DMP update branch then
Philip Durbin 🚀 (Aug 15 2023 at 13:36):I just ran https://github.com/pdurbin/dataverse-api-test-runner/actions/runs/5867681833 against the develop branch and it failed. Perhaps we need a PR merge to develop to fix the develop/unstable images?
Oliver Bertuch (Aug 15 2023 at 13:46):Nope. Someone needs to update https://github.com/IQSS/dataverse/tree/9714-files-api-extension-filters
Oliver Bertuch (Aug 15 2023 at 13:47):It still has the old workflow file because it did not merge develop
Oliver Bertuch (Aug 15 2023 at 13:47):And as it is a branch in the upstream repo, it has access to secrets
Oliver Bertuch (Aug 15 2023 at 13:48):The only way I can think of to prevent this: rename the secret in develop and delete the old - all old workflows will fail that way because no login
Philip Durbin 🚀 (Aug 15 2023 at 14:12):Ah. @Guillermo Portas please take note ^^
Philip Durbin 🚀 (Aug 16 2023 at 10:38):https://github.com/IQSS/dataverse/tree/9714-files-api-extension-filters has been updated
Oliver Bertuch (Aug 23 2023 at 09:53):I think this is solved now. Marking as resolved.
Notification Bot (Aug 23 2023 at 09:53):Oliver Bertuch has marked this topic as resolved.
Last updated: Oct 30 2025 at 05:14 UTC