Dmitry Lysnichenko created AMBARI-6184:
------------------------------------------
Summary: Incorrect value for started_count of Datanode component
Key: AMBARI-6184
URL: https://issues.apache.org/jira/browse/AMBARI-6184
Project: Ambari
Issue Type: Bug
Components: agent
Affects Versions: 1.6.1
Reporter: Dmitry Lysnichenko
Assignee: Dmitry Lysnichenko
Fix For: 1.6.1
*STR:*
# Installed a 3-node cluster for HDP 1.3 stack
HDFS+MapReduce+Nagios+Ganglia+zooKeeper installed with slave components
installed on all 3 hosts.
# Enable security with no kerberos setup
# On expected failure of security wizard, Disable security.
# After successfully disabling security, Following API returns incorrect number
for started_count of Datanode. It says 0 but Datanode is actually running on
all hosts
{code}
http://server:8080/api/v1/clusters/c1/components/?ServiceComponentInfo/category.in(SLAVE,CLIENT)&fields=ServiceComponentInfo/service_name,ServiceComponentInfo/installed_count,ServiceComponentInfo/started_count,ServiceComponentInfo/total_count&minimal_response=true
{code}
Reason:
During wrong kerberos setup DN processes fail to start, but leave stale pid
file owned by root. Next one DN start command starts DN process, but can not
override pid file. So the server considers DN as stopped. If we start DN once
more, commands fail soon after start (due to lock file at data dir owned by
already running DN). Agent reports to server that DN is not running, so server
displays a correct information from his point of view.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
