In T341096: mediawiki-event-enrichment taskmanager crashes at startup we encountered a regression. We deployed in staging and everything was fine, so we proceeded to production. However, in staging, there is no data, so nothing is actually done by the flink app. All that is verified is that the job starts up.

We should include a staging deployment step where we produce data into the kafka cluster / topics that staging uses. At the very least this should be documented at https://wikitech.wikimedia.org/wiki/MediaWiki_Event_Enrichment. Any automation we can provide would help too.