----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/30813/#review71777 -----------------------------------------------------------
What about the "-rollingUpgrade downgrade" option as documented http://hadoop.apache.org/docs/r2.4.0/hadoop-project-dist/hadoop-hdfs/HdfsRollingUpgrade.html. We can make a There's also a rollback option, but I don't think that's going to be supported anytime soon. Should at least investigate the requirements surrounding the downgrade option (also use "query" to see if it's ok to run "downgrade"?). ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/namenode_upgrade.py <https://reviews.apache.org/r/30813/#comment117650> See general comment. We may not ALWAYS be able to "do nothing". - Nate Cole On Feb. 9, 2015, 10:21 p.m., Alejandro Fernandez wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/30813/ > ----------------------------------------------------------- > > (Updated Feb. 9, 2015, 10:21 p.m.) > > > Review request for Ambari, Dmitro Lisnichenko, Jonathan Hurley, Nate Cole, > Tom Beerbower, and Yurii Shylov. > > > Bugs: AMBARI-9541 > https://issues.apache.org/jira/browse/AMBARI-9541 > > > Repository: ambari > > > Description > ------- > > Rolling Upgrade down grade is the same as the upgrade, with the exception > that the "hdfs dfsadmin -rollingUpgrade prepare" command does not need to be > ran. > > > Diffs > ----- > > > ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/namenode_upgrade.py > f74d874 > > ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/params.py > 69cd157 > > ambari-server/src/main/resources/common-services/HDFS/2.1.0.2.0/package/scripts/utils.py > 154e6e3 > > Diff: https://reviews.apache.org/r/30813/diff/ > > > Testing > ------- > > Deployed a 3-node cluster with NamenodeHA, and only HDFS and ZK, then caused > a failure on Datanode, performed a downgrade, and it correctly got to > Finalize. > > Unit tests are in progress. > > > Thanks, > > Alejandro Fernandez > >
