If the errors are only for GlobalVanishJob, then I'd guess that job is setting $params['requestId'] as an integer.

Likely culprit is in this block:

I still think we should do this, even if we are not doing it now.

Hm, actually this looks like it has been happening longer than a week:

Or... you could make a brand new Metrics Platform based instrumentation

oh yes, that would be best!

The aggregate is defined as the collection of all things that are changed together atomically

@Reedy how can we help?

Related: T214998: RFC: Remove m-dot subdomain, serve mobile and desktop variants through the same URL

This should fit into work for WE5 FY2025-2026

Indeed for the upcoming iteration we have T389097 in our current sprint

Can you say more about the reasoning around wanting to make a LogEntryCreated event a subordinate of PageAggregate?

Okay, T389881: Bug: jsonschema-tools generates non deterministic examples for date format fields is done. I also fixed and merged @jsn.sherman's MR 50 in schemas-event-secondary.

@Ottomata in T382147 I went back to just editing the 1.0.0 examples in place instead of incrementing and tests are still failing today;

Patch up for review. I will not have time to write an extensive unit test for this.

Ah ha! I found it.

^ sounds like it will pass until tomorrow!

Hm, you could consider manually editing the 1.0.0 versions and fixing their examples too. Just in case?

Very strange.

^ <3 thank you!

broke those rules ;)

New fields should use the new rules.

You could do both 1. and 2.?

This is a legacy schema, migrated from the older eventlogging system. From https://gitlab.wikimedia.org/repos/data-engineering/schemas-event-secondary

Hm, can you clarify: do you mean to create events for logs? Or to use events to insert into the logging table?

BTW I gave a state of the data platform talk at the WMF Data Strategy convening in November 2024.

If you feel frisky, please submit a patch. The files you'll need to change are static html, at

Related: T258511: Data Lake incremental Data Updates

Oh! incrementing the version is fine too.

Hm, technically you can just update, this only changes the example, so as long as rematerialize all the right files (the fragment and the dependent concrete schemas).

(BTW, maybe the more correct term for this is 'star' rather than 'snowflake'? @JAllemandou ?)

+1 for wmf_content.mediawiki_content_current_v1

We aren't doing any active development work on this, and the plan may change significantly when we do.

Maybe there is a subtle race condition in the test when it dereferences and materializes current.yaml to compare against the materialized file. Somehow maybe it is checking the current.yaml file for examples, not seeing any, generating them, THEN dereferencing and merging schemas together?

I think I see it. I'm not exactly sure why this is happening, but:

If a schema does not have examples, and shouldGenerateExamples is enabled (it is), then https://github.com/json-schema-faker/json-schema-faker is used to generate the examples.

Very strange.

@brouberol lots of patches, is this done?

Thanks Sandra!

BTW, there was a request to do this for varnishkafka, but it was declined when it was intended to do it form ATS instead:

which tried to unify the different link types internally

Duplicate of T386406: Create Gitlab CI templates for JVM packages?

I don't know what is wrong, but some details to help you search:

My question is that whether a bigint or an int field would be "good enough" for your usecases.

@Jdlrobson-WMF @SToyofuku-WMF a couple of your comments made me want to ask a question!

Aye okay, for a minute there it was also reminding me of T309738: Move Mediawiki QueryPages computation to Hadoop.

mass-renaming pages using a maintenance script would emit page moved events, but would not create dummy revisions.

Ideally the system would be generating the data on-the-fly because when permissions change they should ideally be applied immediately.

Status update:
Design doc is in Draft 1.0 state. and is ready for general review.

Ignorant drive by comment:

Note: this task is currently blocking T383814: Upgrade eventgate-wikimedia to node20 as the node 20 upgrade was done after the addition of enrich_fields_with_http_headers.

As this is being considered, please keep in mind that rev_sha1 is used in downstream data pipelines: in the WMF Data Platform, for training ML models, and I'd expect it is used by users of cloud replicas too!

Hi @BCornwall ! group owner approval for analytics-privatedata-users is not needed for WMF or WMDE staff.

FWIW, here is how a move looks in mediawiki.page_change.v1:

Since these are different events, should a listener be reasoning about the order they receive them in?

Huh! For the reconciliation.

Amazing. So we just need to get that bug fixed, convert everything to Iceberg, and then we can stop using JDBC! ;)

Actually, this could be a problem for dumps 2 via page_content_change enrichment! Enrichment happens via a action api uri constructed using meta.domain

referencing the expected database.

Responsible code:

Thought: if we do this, is there a way for the subscriber to submit an event to be processed by a foreign wiki?

By default, streams have canary_events_enabled.
https://wikitech.wikimedia.org/wiki/Data_Platform/Systems/Hadoop_Event_Ingestion_Lifecycle#Canary_Events

Being bold and declining task, please reopen if incorrect.

Being bold and declining this island task.