[ACCUMULO-4197] Intermittent failure of HalfDeadTServerIT

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Cannot Reproduce
• Component/s: test
• Assignee: Unassigned
• Reporter: Josh Elser
• Created: 21/Apr/16 18:12
• Updated: 19/Mar/18 18:13
• Resolved: 19/Mar/18 18:13

Description

I observed an intermittent failure of HalfDeadTServerIT#testRecover today.

This test "injects" pauses to the C read and write calls inside of tserver. testRecover injects a pause of 10 seconds to show that the TServer can reconnect to ZK (and not lose its session).

However, with an RPC timeout of 5 seconds, if a minor compaction is triggered, this pause will cause a compaction to take >10 seconds which will trigger the "hold timeout" causing an exception to be thrown to the client.

2016-04-20 03:39:26,999 [tserver.TabletServer$ThriftClientHandler] ERROR: Commits are held
org.apache.accumulo.tserver.HoldTimeoutException: Commits are held
	at org.apache.accumulo.tserver.TabletServerResourceManager.waitUntilCommitsAreEnabled(TabletServerResourceManager.java:493)
	at org.apache.accumulo.tserver.TabletServer$ThriftClientHandler.flush(TabletServer.java:805)
	at org.apache.accumulo.tserver.TabletServer$ThriftClientHandler.closeUpdate(TabletServer.java:962)
	at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
	at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.lang.reflect.Method.invoke(Method.java:498)
	at org.apache.accumulo.core.trace.wrappers.RpcServerInvocationHandler.invoke(RpcServerInvocationHandler.java:46)
	at org.apache.accumulo.server.rpc.RpcWrapper$1.invoke(RpcWrapper.java:73)
	at com.sun.proxy.$Proxy21.closeUpdate(Unknown Source)
	at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor$closeUpdate.getResult(TabletClientService.java:2446)
	at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor$closeUpdate.getResult(TabletClientService.java:2430)
	at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:39)
	at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:39)
	at org.apache.accumulo.server.rpc.TimedProcessor.process(TimedProcessor.java:63)
	at org.apache.thrift.server.AbstractNonblockingServer$FrameBuffer.invoke(AbstractNonblockingServer.java:516)
	at org.apache.accumulo.server.rpc.CustomNonBlockingServer$1.run(CustomNonBlockingServer.java:78)
	at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
	at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
	at org.apache.accumulo.fate.util.LoggingRunnable.run(LoggingRunnable.java:35)
	at java.lang.Thread.run(Thread.java:745)

I believe this is intermittent based on whether or not a minor compaction is triggered after the i/o pausing is triggered or not. We could probably work around this by increasing the threshold for running a minor compaction.