I noticed in your tserver debug logs: 2014-03-03 14:54:05,231 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:05,482 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:05,733 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:05,984 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:06,234 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:06,485 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:06,736 [util.TServerUtils] INFO : Unable to use port 9997, retrying. (Thread Name = Thrift Client Server) 2014-03-03 14:54:07,005 [metrics.MetricsConfiguration] DEBUG: Loading config file: /opt/accumulo/accumulo/conf/accumulo-metrics.xml 2014-03-03 14:54:07,019 [metrics.MetricsConfiguration] DEBUG: Metrics collection enabled=false 2014-03-03 14:54:07,025 [tabletserver.TabletServer] INFO : port = 9997 2014-03-03 14:54:07,080 [tabletserver.TabletServer] INFO : Waiting for tablet server lock
Are you using Cloudera Manager for HDFS+Mapred+Zookeeper? I ask because the Service Monitor Listen Port for the Cloudera Agent uses port 9997 by default. It will create a conflict if already in use by Cloudera Manager. Also a, $ netstat –taupen | grep 9997 Should tell you what PID is using that port. Maybe you already have an old tablet server running that needs to be killed? From: [email protected] [mailto:[email protected]] On Behalf Of Alex Lee Sent: Wednesday, March 05, 2014 1:04 PM To: [email protected] Subject: RE: Tablet server stuck waiting for lock Here are links to the logs: Master https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/master/master_u2acctm.log https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/master/master_u2acctm.debug.log Node 1 https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/node1/tserver_u2x1acc1.log https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/node1/tserver_u2x1acc1.debug.log Node 2 https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/node2/tserver_u2x2acc1.log https://dl.dropboxusercontent.com/u/20780006/Accumulo%20logs/node2/tserver_u2x2acc1.debug.log Alex From: Sean Busbey [mailto:[email protected]] Sent: Wednesday, March 05, 2014 12:46 PM To: Accumulo User List Subject: Re: Tablet server stuck waiting for lock On Wed, Mar 5, 2014 at 11:36 AM, Alex Lee <[email protected]<mailto:[email protected]>> wrote: Thank you for the quick response. Where should I post the logs for you to access them? I’m relatively new to mailing lists, so I’m not sure what standard procedure for that is. There isn't really a standard procedure, it varies based on how big they are and what kind of resourcing you have. The easiest thing is if you have somewhere you can host them yourself, like a file share. If they're not too big, you could use a paste bucket, i.e the Apache one[1]. [1]: community: http://apaste.info/ committers: https://paste.apache.org/
