Details
-
Bug
-
Resolution: Fixed
-
Medium
-
4.13.0
-
3
-
Severity 1 - Critical
-
45
-
Description
Issue Summary
When SLA configuration changes all issues in the project may be re-indexed.
The re-index is done in a heavy way (real-time cross-cluster re-index) and can affect an unlimited number of issues.
This can cause the whole instance to fail.
Steps to Reproduce
- Create a project with >50K issues on a multinode DC instance
- Change SLA configuration in the project
Expected Results
SLAs on the project should be updated and no performance impact
Actual Results
A huge number of indexing calls cause high CPU load and slow down the whole instance. Other indexing task may start to timeout (DBR - real time index replication, NodeReindexService - delayed node index replication/index consistency check, ... any other re-indexing triggered by background, ie. non-http threads). This may cause index inconsistency between nodes which can eventually be only resolved with a full foreground reindex.
Workaround
Currently there is no workaround for the problem.
Attachments
Issue Links
- caused
-
HOT-96659 Loading...
- mentioned in
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
- PIR - Medium Priority Action
-
- relates to
-