we should find out why the message to 'initialize traf, drop' was returned. Have we done any incompatible metadata changes?
Or is the error due to inaccessible hbase when sqlci comes up? At sqlci startup time, we try to access hbase to read couple of metadata tables using hbase apis. It is possible that somehow hbase error is being reported as init drop error. anoop -----Original Message----- From: Roberta Marton [mailto:[email protected]] Sent: Tuesday, February 9, 2016 12:17 PM To: [email protected] Subject: RE: initialize trafodion after upgrade I had the same issue yesterday but was not sure if it was my problem (I am fixing some data corruption issues). Anyway, I ended up uninstalling and reinstalling the Hadoop eco-system to get things to work again. Also rebuilt everything. Hopefully, there is a better solution. Roberta -----Original Message----- From: Dave Birdsall [mailto:[email protected]] Sent: Tuesday, February 9, 2016 12:03 PM To: [email protected] Subject: initialize trafodion after upgrade Hi, This morning I brought an existing development instance up-to-date (did the git fetch –all, and created a new branch, did an sqstop, make clean, make all). When I brought the instance back up, it complained that I needed to do an initialize trafodion, drop + an initialize trafodion. I did the initialize trafodion, drop without problems. But the initialize trafodion fails as follows: Apache Trafodion Conversational Interface 2.0.0 Copyright (c) 2015 Apache Software Foundation >>initialize trafodion; *** ERROR[8448] Unable to access Hbase interface. Call to ExpHbaseInterface::create() returned error HBASE_CREATE_ERROR(701). Cause: java.lang.NullPointerException java.lang.String.contains(String.java:2076) org.trafodion.sql.CoprocessorUtils.addCoprocessor(CoprocessorUtils.java:76) org.trafodion.sql.HBaseClient.createk(HBaseClient.java:490) . --- SQL operation failed with errors. >>exit; So, I’m guessing the coprocessor jars need to be brought up-to-date. Is there a convenient script to do this? Or must I bite the bullet and do a swuninstall + install_local_hadoop? Thanks, Dave
