Details
-
Bug
-
Status: Open
-
Minor
-
Resolution: Unresolved
-
3.0.0-alpha1
-
None
-
None
Description
We currently have the following race:
- Block is underreplicated, hence it's present in neededReplications
- User deletes the block - its BlockInfo.blockCollection is set to null
- Admin runs metaSave before the replication monitor runs.
This causes an NPE since block.getBlockCollection() for one of the neededReplication blocks has become null.
Attachments
Issue Links
- duplicates
-
HDFS-4867 metaSave NPEs when there are invalid blocks in repl queue.
- Closed