----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/30912/#review72145 -----------------------------------------------------------
Ship it! Ship It! - Dmitro Lisnichenko On Feb. 12, 2015, 2:37 a.m., Alejandro Fernandez wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/30912/ > ----------------------------------------------------------- > > (Updated Feb. 12, 2015, 2:37 a.m.) > > > Review request for Ambari, Dmitro Lisnichenko, Jonathan Hurley, Nate Cole, > Robert Levas, and Tom Beerbower. > > > Bugs: AMBARI-9559 > https://issues.apache.org/jira/browse/AMBARI-9559 > > > Repository: ambari > > > Description > ------- > > I reopened AMBARI-9559 for this additional patch. > > While performing a RU, it failed in Finalize because the SECONDARY_NAMENODE > component was still on an old version, since it was not deleted by the > Namenode HA wizard. > This was a 3-node cluster with HDFS, MR, YARN, TEZ, Hive, ZK, Pig, AMS using > HDP 2.2.0.0 GA. > > I was not able to reproduce this, even after installing the exact same > services on the same hosts, and using the same ambari-server version. It > could be the case that some thread is holding a reference to the object that > is being deleted, so it persists. So I added another check to > PreUpgradeCheckResourceProvider that ensures that SECONDARY_NAMENODE is not > present on any hosts. > > > Diffs > ----- > > > ambari-server/src/main/java/org/apache/ambari/server/checks/SecondaryNamenodeDeletedCheck.java > ad967f8 > > Diff: https://reviews.apache.org/r/30912/diff/ > > > Testing > ------- > > Deployed a 3-node cluster with HDFS, ZK, and added Namenode HA. I was able to > reproduce SECONDARY_NAMENODE not being deleted. > I then registered a repo and distributed it to the hosts, and when I tried to > perform a RU, the precheck correctly reported an error. > > Unit tests are in progress. > > > Thanks, > > Alejandro Fernandez > >
