Details
-
Sub-task
-
Status: Resolved
-
Critical
-
Resolution: Not A Problem
-
2.0.0-alpha, 0.23.6, 3.0.0-alpha1
-
None
-
None
Description
The RM stops trying to renew tokens if any exception occurs during the renew. This should be changed to abort only if the exception is InvalidToken to allow resilience to transient network failures, issues associated with aborted connections when the NN is overloaded, cluster upgrades, etc.
Attachments
Issue Links
- relates to
-
YARN-2836 RM behaviour on token renewal failures is broken
- Open