ElasticSearch Repair
On rare occasions, ElasticSearch may fall out of sync. The most likely example is becoming out of sync with Cassandra after an unexpected system down event, and typically it would recover itself in no more than 24 hours.
If you are experiencing data inconsistency issues, for example – people or channels not appearing in search results, or some users' online/offline status indicators do not match their true values – then it may be helpful to execute the following re-sync procedure:
Note: In a Georedundant setup, the following procedure should be performed in each data center.
kubectl \
-n ice-arcus \
create job \
resync-$(date "+%Y%m%d-%H%M") \
--from=cronjob/elastic-sync-DATACENTERNAME