Eric, I had previously done a stop-all.sh, so I started everything back up again to check this.
Zoo1:2181, follower, 4 clients Zoo2:2181, leader, 3 clients Zoo3:2181, follower, 5 clients The behavior of the tablet servers seems to be inconsistent. Upon restarting Accumulo, the overview page gives the impression that everything is fine. However, it is only listing 1 tablet server (Node 1), and it is now tablet server #2 that is timing out while "Waiting for tablet server lock". There are no errors in tablet server #1's logs right now, while tablet server #2 failed after too many retries of waiting for the lock. Also, the master log still has no mention of the second node's IP address, even though the start-all.sh script indicates that the tserver and logger are being started on both nodes. Thanks, Alex From: Eric Newton [mailto:[email protected]] Sent: Wednesday, March 05, 2014 1:25 PM To: [email protected] Subject: Re: Tablet server stuck waiting for lock On the monitor page, there's a box that shows your zookeepers and their status. What does it say? -Eric On Wed, Mar 5, 2014 at 1:09 PM, Alex Lee <[email protected]<mailto:[email protected]>> wrote: Dfs permissions is currently disabled. I'm using the accumulo user for "accumulo init" and for "start-all.sh", and it is also the user that has passwordless SSH enabled. I ran "hadoop fs -ls /accumulo" as the accumulo user on both tablet servers, and I am able to see inside of the /accumulo directory on hdfs. Alex From: Ott, Charlie H. [mailto:[email protected]<mailto:[email protected]>] Sent: Wednesday, March 05, 2014 1:02 PM To: [email protected]<mailto:[email protected]> Subject: RE: Tablet server stuck waiting for lock The connection reset by peer from the Master in combination with the lock not acquired by the tablet server makes me wonder if the process owner for the tablet server is able to access HDFS correctly. Are dfs permissions enabled on your HDFS? It makes me think the tablet server does not have permissions to read from the /accumulo path that was initialized on the master. Did you use the same account for 'accumulo init' ? From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Alex Lee Sent: Wednesday, March 05, 2014 12:17 PM To: [email protected]<mailto:[email protected]> Subject: Tablet server stuck waiting for lock Hello, I'm trying to create a virtualized Accumulo 1.4.4 cluster with 4 tablet servers using Hadoop 0.20.2 and ZooKeeper 3.3.5. It didn't seem to be working correctly with 4 tablet servers, so I first tried just running with one tablet server, which seemed to work fine. When I tried to run it with just 2 tablet servers, I ran into some issues. Just to preface, I double checked configs within zookeeper and accumulo, and everything matches. All hostnames are resolving correctly, and passwordless SSH for the accumulo user is also functional between all nodes. Running "echo stat | nc <zk-server> <zk port>" responds appropriately. Here's the first error log for the Tablet Master: 2014-03-05 11:18:16,626 [master.Master] ERROR: Error processing table state for store Root Tablet org.apache.thrift.transport.TTransportException: java.io.IOException: Connection reset by peer at org.apache.thrift.transport.TIOStreamTransport.flush(TIOStreamTransport.java:161) at org.apache.thrift.transport.TFramedTransport.flush(TFramedTransport.java:158) at org.apache.accumulo.core.client.impl.ThriftTransportPool$CachedTTransport.flush(ThriftTransportPool.java:299) at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Client.send_loadTablet(TabletClientService.java:653) at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Client.loadTablet(TabletClientService.java:640) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source) at java.lang.reflect.Method.invoke(Unknown Source) at org.apache.accumulo.cloudtrace.instrument.thrift.TraceWrap$2.invoke(TraceWrap.java:84) at com.sun.proxy.$Proxy4.loadTablet(Unknown Source) at org.apache.accumulo.server.master.LiveTServerSet$TServerConnection.assignTablet(LiveTServerSet.java:86) at org.apache.accumulo.server.master.Master$TabletGroupWatcher.flushChanges(Master.java:1818) at org.apache.accumulo.server.master.Master$TabletGroupWatcher.run(Master.java:1426) Caused by: java.io.IOException: Connection reset by peer at sun.nio.ch.FileDispatcherImpl.write0(Native Method) at sun.nio.ch.SocketDispatcher.write(Unknown Source) at sun.nio.ch.IOUtil.writeFromNativeBuffer(Unknown Source) at sun.nio.ch.IOUtil.write(Unknown Source) at sun.nio.ch.SocketChannelImpl.write(Unknown Source) at org.apache.hadoop.net.SocketOutputStream$Writer.performIO(SocketOutputStream.java:55) at org.apache.hadoop.net.SocketIOWithTimeout.doIO(SocketIOWithTimeout.java:142) at org.apache.hadoop.net.SocketOutputStream.write(SocketOutputStream.java:146) at org.apache.hadoop.net.SocketOutputStream.write(SocketOutputStream.java:107) at java.io.BufferedOutputStream.flushBuffer(Unknown Source) at java.io.BufferedOutputStream.flush(Unknown Source) at org.apache.thrift.transport.TIOStreamTransport.flush(TIOStreamTransport.java:159) ... 13 more Here are the error logs for Tablet Server #1: 2014-03-05 11:17:15,152 [tabletserver.TabletServer] INFO : Tablet server starting on 172.16.111.3 2014-03-05 11:17:15,187 [util.FileSystemMonitor] INFO : Filesystem monitor started 2014-03-05 11:17:15,194 [tabletserver.NativeMap] INFO : Loaded native map shared library /opt/accumulo/accumulo/lib/native/map/libNativeMap-Linux-amd64-64.so 2014-03-05 11:17:15,499 [tabletserver.TabletServer] INFO : port = 9997 2014-03-05 11:17:15,540 [tabletserver.TabletServer] INFO : Waiting for tablet server lock 2014-03-05 11:17:16,633 [tabletserver.TabletServer] WARN : Got loadTablet message from master before lock acquired, ignoring... 2014-03-05 11:17:16,634 [server.TNonblockingServer] ERROR: Unexpected exception while invoking! java.lang.RuntimeException: Lock not acquired at org.apache.accumulo.server.tabletserver.TabletServer$ThriftClientHandler.checkPermission(TabletServer.java:1782) at org.apache.accumulo.server.tabletserver.TabletServer$ThriftClientHandler.loadTablet(TabletServer.java:1814) at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method) at sun.reflect.NativeMethodAccessorImpl.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(Unknown Source) at java.lang.reflect.Method.invoke(Unknown Source) at org.apache.accumulo.cloudtrace.instrument.thrift.TraceWrap$1.invoke(TraceWrap.java:59) at com.sun.proxy.$Proxy1.loadTablet(Unknown Source) at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor$loadTablet.process(TabletClientService.java:2510) at org.apache.accumulo.core.tabletserver.thrift.TabletClientService$Processor.process(TabletClientService.java:2053) at org.apache.accumulo.server.util.TServerUtils$TimedProcessor.process(TServerUtils.java:154) at org.apache.thrift.server.TNonblockingServer$FrameBuffer.invoke(TNonblockingServer.java:631) at org.apache.accumulo.server.util.TServerUtils$THsHaServer$Invocation.run(TServerUtils.java:202) at java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source) at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source) at org.apache.accumulo.core.util.LoggingRunnable.run(LoggingRunnable.java:34) at java.lang.Thread.run(Unknown Source) 2014-03-05 11:17:20,564 [tabletserver.TabletServer] INFO : Waiting for tablet server lock 2014-03-05 11:17:25,589 [tabletserver.TabletServer] INFO : Waiting for tablet server lock (continues until too many retries, then exits) Tablet Server #2's logs get as far as this (below), and then just stop. 2014-03-05 11:17:14,112 [tabletserver.TabletServer] INFO : Tablet server starting on 172.16.111.3 2014-03-05 11:17:14,149 [util.FileSystemMonitor] INFO : Filesystem monitor started 2014-03-05 11:17:14,157 [tabletserver.NativeMap] INFO : Loaded native map shared library /opt/accumulo/accumulo/lib/native/map/libNativeMap-Linux-amd64-64.so 2014-03-05 11:17:14,481 [tabletserver.TabletServer] INFO : port = 9997 Also, the master logs interestingly never make any calls to Tablet #2's IP address. Any thoughts? We have another cluster that is setup identically in just about every way (besides hostnames), but it has never experienced any of these issues. My research shows that these issues can exist within 1.4.3, which we were using at first, but we switched to 1.4.4 because these types of issues were supposed to be resolved. Any help would be greatly appreciated. Thanks, Alex Lee
