That will be fixed in the next version of Ambari. https://issues.apache.org/jira/browse/AMBARI-13713
From: Brian Jeltema <[email protected]<mailto:[email protected]>> Reply-To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Date: Wednesday, November 18, 2015 at 10:08 AM To: "[email protected]<mailto:[email protected]>" <[email protected]<mailto:[email protected]>> Subject: Re: tez failure after upgrade I was running Ambari 2.1.2.1. The problem was that the value of tez.lib.uris was not updated by the rolling upgrade. Updating it and restarting the tez clients fixed the problem. Thanks. Brian On Nov 18, 2015, at 12:43 PM, Hitesh Shah <[email protected]<mailto:[email protected]>> wrote: Hello Brian, What version of Ambari are you running? I believe there were bugs in earlier versions that uploaded the wrong version of the Tez tarball to HDFS during an upgrade. You can check this by comparing the tez tarball on HDFS ( pointed to by tez.lib.uris in tez-site.xml ) to the tez tarball in the respective version of the Tez rpm. This can be easily fixed by deleting the wrong tarball in HDFS under the the 2.3.2 version dir and replacing it with the correct one. This may also impact the MapReduce tarball too but MR jobs do not fail as they don't have a compatibility check in place. thanks - Hitesh On Nov 18, 2015, at 8:58 AM, Brian Jeltema <[email protected]<mailto:[email protected]>> wrote: On completion of a rolling upgrade to HDP 2.3, tez failed it's service check with: 15/11/18 11:44:49 INFO client.DAGClientImpl: DAG completed. FinalState=FAILED 15/11/18 11:44:49 INFO examples.OrderedWordCount: DAG diagnostics: [Incompatible versions found, clientVersion=0.7.0.2.3.2.0-2950, AMVersion=0.5.2.2.2.8.0-3150] The clientVersion is the version after the upgrade, but the AMVersion is the version before the upgrade. Does anyone know what might cause this? Thanks Brian
