Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
These changes are from a recent high-concurrency test-run with
- 4x HS2 (each with dedicated HMS instances), each with up to 2k sessions
- 1x HS2 (with dedicated HMS instance) for housekeeping (~all background tasks)
The tests brought two issues to the surface:
- missing indexes caused long running queries in the backend database (with table scans) which then lead to query locks
- the notification log cleaners can't be configured to a single node
