Found the error in the logs when this happened: good news/bad news

@EBomani had the suspicion that this has to do with kernel inotify limits.

In T383243#10445830, @dduvall wrote:

Since trees, like all Git objects, are content addressable, two refs that result in the same working tree checkout should be referencing the same exact tree. Right? (Serious question.)

Tagging in @ArthurTaylor and @kostajh since this idea seems similar to work they're exploring in parallelizing tests within a single test suite.

In T706#10438801, @Seddon wrote:

Hey I have two batch requests on behalf of the apps engineering group.

iOS Team: Can you add @HNordeenWMF, @Tsevener, @Mazevedo and @GOlson-WMF to acl*Project-Admins? These folks are members of the iOS team and support rotating sprint responsibilities and need the ability to fully manage the Wikipedia-iOS-App-Backlog board and other phab tags as the team starts adopting formal sprints.

Android Team: Can you also add @cooltey and @WRai-WMF to acl*Project-Admins? These folks are members of the Android team and need the ability to fully manage the Wikipedia-Android-App-Backlog board and other phab tags as the team starts adopting formal sprints.

Thanks for the poke @Dzahn

In T383083#10434158, @HCoplin-WMF wrote:

Original image dimensions were 4672 × 5941, 3.8 MB (https://phabricator.wikimedia.org/F58028490). I then trimmed it down to be more square, resulting in 4667x4763, 4.3 MB (https://phabricator.wikimedia.org/F58028494). Both of those successfully uploaded, but displayed the mystery ghost.

The final image that worked was 400x400, 91 KB (https://phabricator.wikimedia.org/F58134690).

It seems like the dimensions of the first images were within the ratio -- was the file size too big, even though it allowed the upload?

Other idea: we could have some bot go around and move self-service archived repos into All-Archived-Projects ¯\_(ツ)_/¯. Best of both worlds: people continue to archive, Gerrit doesn't waste time replicating.

These kinds of things will get filed as Wikimedia-production-error tasks (usually) once a patch has merged, so they might be useful in PatchDemo environments. Declining for now.

This would involve adding a step to the test pipeline in zuul/layout.yaml. And defining a test to run somewhere in JJB within that same repo.

De-prioritizing while discussing other options of where Catalyst might run in future. Keeping the task around in the backlog in case it's still useful going forward.

In T706#10419719, @VPuffetMichel wrote:

Hi!

I'd like to become project admin for the "editing team" to be able to create milestones https://phabricator.wikimedia.org/project/profile/2823/

and also for the campaigns team.
Thank you so much!

Last train 2024 is now complete!

In T381851#10401485, @Scott_French wrote:

@Ammarpad - FYI, @thcipriani is out this week, so the next update here will likely be next week. Thanks for your patience.

Pausing for holiday break

For clarity, as you noticed: these tokens are useless by the time the image is published. I wonder if the presence in the .git dir is gitlab's doing vs kokkuri's? @dancy or @dduvall would know for sure.

In T365070#9856112, @sbassett wrote:

Access to security issues works just as fine, we can monitor for anything that impacts wiki.gg. Sorry for the confusion on my end.

So volunteer security bug access is unfortunately a very difficult issue. We only have a handful of folks who have been given this access and they are typically former or current WMF staff and/or are extremely active in triaging and working on security issues within the Wikimedia ecosystem. We've tried offering more limited security bug access in the past to various MediaWiki operators (e.g. T101017) but that wasn't possible since it's all-or-nothing access to every security bug within Phabricator, including some extremely sensitive issues regarding current and past security incidents. I'm not going to say that this isn't possible, but it would require WMF-Legal and various leadership at the WMF understanding and accepting this risk.

In T380525#10357449, @elukey wrote:

@thcipriani Hi! I'd need your review to grant access to the Deployment group, lemme know your thoughts :)

In T381408#10376934, @andrea.denisse wrote:

Hi @thcipriani do you approve this request?

In T381123#10370380, @elukey wrote:

@thcipriani Hi! Could you review this request? Lemme know if we can move forward :)

Team offsite this week, no train.

There is an existing multi-node k3s in the catalyst-qte project. Still TODO benchmarking storage solutions (an added solution to test is using daemonsets for local path provisioners).

Resolved pending deploy by @EBomani

Resolved pending deploy by @EBomani

Deployed and merged (not in that order)

Resolved pending deploy by @EBomani

Noting that this week's train is happening over US Thanksgiving. Due to the holiday, if we run into a blocker we need the US cabal to resolve, then we may need to abandon this week's train and pick it up with 1.44.0-wmf.6 next week.

In T380535#10346862, @thcipriani wrote:

A strange datapoint here:

~~~https://gerrit.wikimedia.org/g/ shows design/codex-php but not its description. But it clearly has a description.~~~ nevermind. It had no description.

A strange datapoint here:

Adding @cscott per Developers/Maintainers (update page if incorrect, please and thank you!)