[HIVE-17758] NOTIFICATION_SEQUENCE_LOCK_RETRY_SLEEP_INTERVAL.defaultLongVal is -1

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.0.0
• Assignee: Zoltan Haindrich
• Reporter: Zoltan Haindrich
• Created: 10/Oct/17 12:33
• Updated: 22/May/18 23:16
• Resolved: 11/Oct/17 15:15

Description

HIVE-16886 introduced a retry logic; which has a configurable retry interval. unfortunately HiveConf has some public fields which at first glance seems to be usefull to pass as arguments to other methods - but in this case the default value is not even loaded into the field read by the code.. and because of that the innocent client code here have used a -1 value incorrectly which eventually caused an exception here:

2017-10-10 11:22:37,638 ERROR [load-dynamic-partitions-12]: metastore.ObjectStore (ObjectStore.java:addNotificationEvent(7444)) - could not get lock for update
java.lang.IllegalArgumentException: timeout value is negative
        at java.lang.Thread.sleep(Native Method)
        at org.apache.hadoop.hive.metastore.ObjectStore$RetryingExecutor.run(ObjectStore.java:7407)
        at org.apache.hadoop.hive.metastore.ObjectStore.lockForUpdate(ObjectStore.java:7361)
        at org.apache.hadoop.hive.metastore.ObjectStore.addNotificationEvent(ObjectStore.java:7424)
        at sun.reflect.GeneratedMethodAccessor71.invoke(Unknown Source)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:498)
[...]

Issue Links

• is related to: HIVE-17759 Prevent the misuses of HiveConf.ConfVars.default fields
• relates to: HIVE-16886 HMS log notifications may have duplicated event IDs if multiple HMS are running concurrently