Page MenuHomePhabricator

All events in the contenttranslationabusefilter data stream failing validation
Closed, ResolvedPublic


All the contenttranslationabusefilter events appear to be failing validation because '.event.filterId' should be integer (see this Logstash view). It seems the client is submitting a string instead.

The event.contenttranslationabusefilter Hive table has no data at all, meaning the error was introduced more than 90 days ago.

Event Timeline

Tagging Analytics and Product-Data-Infrastructure for their awareness as this might be a result of migrating the stream to the Event Platform. At any rate, it should be an easy fix for the Language team.

Change 699399 had a related patch set uploaded (by Nik Gkountas; author: Nik Gkountas):

[mediawiki/extensions/ContentTranslation@master] CX eventlogging: Fix ContentTranslationAbuseFilter filterId

Change 699399 merged by jenkins-bot:

[mediawiki/extensions/ContentTranslation@master] CX eventlogging: Fix ContentTranslationAbuseFilter filterId

I think this is fixed now.

Logstash now shows no validation errors since the fix was deployed.

The Hive database is now receiving a healthy number of events:

year    month   day     events
2021	6	24	103
2021	6	25	561
2021	6	26	1141
2021	6	27	1260
2021	6	28	613
2021	6	29	162

The dashboard seems to be up and running again. Thanks!

superset.wikimedia.org_superset_dashboard_cx-abuse-filter_(iPad).png (1×2 px, 218 KB)