Profiling

[sadatmalik]

• [x] Confirm Ehcaching is working
• [x] Implement cache eviction
• [x] Id, email, phone number query related indexing
• [ ] CandidateOpportuntyRepository opp/check-unread-chat
• [x] Understand cpu profile discrepancy between AWS and Jprofiler
• [x] Open an AWS support ticket
• [x] Consider using AWS cloud profiler?
• [x] Investigate cpu spikes post start-up
• [ ] Document JProfiler steps on TC Wiki
• [ ] Team JProfiler licenses
• [ ] Break out JobChats micro service?

---

• Introduced a 30-minute max lifetime for prod DB connections in our Hikari config to alleviate stale DB connections as a Hotfix. This trumps the 24-hour timeout setting on the Postgres RDS DB.

• A PR for the same config is prepped and ready for staging.

• Task 24 in Staging is created for deployment once the above PR merges.

• Rebooted prod DB and despite seeing no stale connections, still have high CPU task usage & lingering sluggishness.

• Boosted AWS CPU to 1024 from 512, this has reduced sluggishness a little but still seeing high CPU spikes.

• Plan to dial back to 512 CPU once we unravel the mystery behind the high CPU usage. (edited)

[sadatmalik]

Next avenues to explore - CandidateOpportuntyRepository opp/check-unread-chats:

[samschlicht]

Closed issue #794 is worth taking a look at again if we notice any connection between CPU spikes and file uploads.