Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
1.0.0
-
None
-
None
Description
When lock acquisition times out, it would be useful to include info in the message about the lock that caused the current request to block.
It will help identify runaway processes, etc.
This would require a Thrift change to pass that info up to the client which determines when to give up waiting.
Implementation:
In case WAIT: in TxnHandler.checkLock(Connection dbConn, long extLockId) add to org.apache.hadoop.hive.metastore.api.LockResponse info from
locks[i]
which already has the ids and human readable info about conflicting lock.
Attachments
Issue Links
- blocks
-
HIVE-12529 HiveTxnManager.acquireLocks() should not block forever
- Closed