Details
-
Sub-task
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
None
-
None
-
None
-
None
Description
Currently there is a config (10ms) which can be an issue for large clusters with many jobs heartbeating to the RM. We should be able to scale it up and down based on outstanding requests etc. e.g. if there are no outstanding requests then ping on larger intervals. The heuristics may be more sophisticated than that.
Attachments
Issue Links
- duplicates
-
TEZ-119 The AM-RM heartbeat interval should not be static
- Reopened