[HADOOP-12950] ShutdownHookManager should have a timeout for each of the Registered shutdown hook

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.8.0, 3.0.0-alpha1
• Hadoop Flags: Reviewed
• Assignee: Xiaoyu Yao
• Reporter: Xiaoyu Yao
• Created: 21/Mar/16 20:01
• Updated: 22/Aug/18 00:57
• Resolved: 31/Mar/16 22:25

Description

HADOOP-8325 added a ShutdownHookManager to be used by different components instead of the JVM shutdownhook. For each of the shutdown hook registered, we currently don't have an upper bound for its execution time. We have seen namenode failed to shutdown completely (waiting for shutdown hook to finish after failover) for a long period of time, which breaks the namenode high availability scenarios. This ticket is opened to allow specifying a timeout value for the registered shutdown hook.

Issue Links

• causes: HADOOP-15679 ShutdownHookManager shutdown time needs to be configurable & extended
• is related to: HADOOP-12993 Change ShutdownHookManger complete shutdown log from INFO to DEBUG
• is related to: SPARK-25183 Spark HiveServer2 registers shutdown hook with JVM, not ShutdownHookManager; race conditions can arise
• is related to: HADOOP-8325 Add a ShutdownHookManager to be used by different components instead of the JVM shutdownhook
• is related to: HADOOP-4829 Allow FileSystem shutdown hook to be disabled
• relates to: HADOOP-10219 ipc.Client.setupIOstreams() needs to check for ClientCache.stopClient requested shutdowns