Remove search indexes when an object is destroyed. This was hinted at by a TODO comment, and generally seems reasonable.
Details
Details
- Reviewers
epriestley - Group Reviewers
Blessed Reviewers
- Set search.elastic.host and ran ./bin/search init and ./bin/search index --all.
- At this point the search index exists in both MySQL and ElasticSearch.
- Removed an object with ./bin/remove destroy.
- Queried phabricator_search.search_document for the removed PHID.
- Queried ElasticSearch with curl -XGET http://elasticsearch:9200/phabricator/TASK/PHID-TASK-7uv6xhcexixtroqvssqw/.
Diff Detail
Diff Detail
- Repository
- rP Phabricator
- Branch
- master
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 6538 Build 6560: [Placeholder Plan] Wait for 30 Seconds
Event Timeline
Comment Actions
Not really related to this diff... but maybe object destruction should be queued for taskmaster daemons instead of handled synchronously.
Comment Actions
This stuff should be implemented as a DestructionEngineExtension now, but seems basically correct/reasonable to me.
The extension can test for $object instanceof PhabricatorFulltextInterface to detect whether or not it may have fulltext storage.