Description
Task history currently does not deliver any retention guarantees beyond a max of 100 tasks per job over at most 2 days (whichever comes first). This makes any immediate event processing (i.e. SLA monitoring) impossible under heavy turn over.
Have the HistoryPrunner support a min configurable retention period (e.g. 2 minutes) to mitigate the above problem. The risk of overloading mem store is minimal due to throttling of flapping tasks.
Attachments
Issue Links
- is related to
-
AURORA-261 on sandbox garbage collection, just de-link the slave HREF on the finished tasks page
- Resolved