To complete the migration to k8s we need to work on enabling all of the changeprop tasks that we currently do in production to bring the two instance types into parity.
While these features are being rolled out, we should develop and document test data for these features.
Before we can roll tasks out, we should
- implement feature toggles in the helm chart to enable easier rollout and reduce the need for continuous repackaging
Tasks that need to be enabled:
- mobile-html_rerender_transclude
- mobile-html_rerender
- media-list_rerender_transcludes
- media-list_rerender
- mobile-sections_rerender_transcludes
- mobile-sections_rerender
- summary_definition_rerender
- summary_definition_rerender_transcludes
- metadata_purge
- metadata_purge_transcludes
- mw_purge
- purge_varnish - k8s in prod now produces events to another topic. This currently relies on the scb changeprop instance but as far as kubernetes migration is concerned this is complete
- null_edit
- page_edit
- revision_visibility_change
- page_delete
- page_restore
- page_move
- on_transclusion_update
- page_create
- on_backlinks_update
- ores_cache
- wikidata_description_on_edit
- wikidata_description_on_undelete
- on_wikidata_description_change
- page_images_summary
- page_images_mobile
We also need to
- build monitoring dashboards for the new instances that complement or mimic the existing dashboards we have
Finally when ready we need to
- decommission existing changeprop scb instances- blocked on revision of purge mechanism