I made some changes in my cluster and I set up Kerberos, after restarted all services, the Resource manager would not start, this is what I have in My log file:
Service org.apache.hadoop.yarn.server.resourcemanager.EmbeddedElectorService failed in state INITED; cause: java.io.IOException: Couldn't create /yarn-leader-election
I try to understand what going wrong. I found this question that is similar to mine: https://community.hortonworks.com/questions/71752/resourcemanagersha-dont-start.html
BTW, I have my Resource manager installed on 2 masters nodes.
How could I fix this Problem
any suggestion would be greatly appreciated
Did you check the ACL as mentioned in the other question?
/usr/hdp/current/zookeeper/bin/zkCli.sh -server 127.0.0.1:2181 [zkshell] getAcl /
the 'cdrwa' permission should be fine, 'r' is readonly. Please check if the path /yarn-leader-election exists:
[zkshell] ls /yarn-leader-election [zkshell] getAcl /yarn-leader-election
and if it exists, you might simply try to delete the zookeeper pat and try a restart of yarn:
[zkshell] delete /yarn-leader-election
if the path does not exist, or yarn doesn't start even after deleting it, we will have to get deeper.