Details
-
Sub-task
-
Status: Open
-
Minor
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
In big cluster, if one of DN do not respond within 5min, due to slow running or other cases,
- SCM will mark it as stale and closes pipeline and all containers
- If DN response is received, again SCM will create all pipeline and cotainers
This creates lot of event, extra burden over SCM and DN.
Other impact:
- Few created container remains empty or small container
To handle the ripple effect of closure of pipeline and containers and re-creation,
- Can mark pipeline as readonly, i.e. it will not be used further for any new block allocation
- Containers/pipeline will remain open, till either DN becomes healthy, or DN is dead