Description of problem: RHQ Controll - rhqctl stop --agent removed agent.pid but doesn't stop process Version-Release number of selected component (if applicable): jenkins build 177 How reproducible: always Steps to Reproduce: 1. run ./rhqctl install --storage 2. run ./rhqctl stop --agent 3. run ./rhqctl start --agent Actual results: After step 1 both agent and storage are installed and running After step 2 agent process is running while the pid file is removed After step 3 no new agent is started - log is "INFO [org.jboss.modules] JBoss Modules version 1.1.1.GA" Expected results: After step 1 both agent and storage are installed and running After step 2 agent process is stopped and the pid file is removed After step 3 agent is started - RHQ Agent (pid {number} running" message. Additional info: http://jenkins.jonqe.lab.eng.bos.redhat.com:9080/job/RHQ_Control_Run/44/consoleFull -- please search for rhqctl start --agent
I am seeing this issue consistently. I tested a master build, and did not see the issue there. I think the problem is in the Cassandra plugin. It uses the Cassandra CQL driver which which uses its own internal thread pool. I think that the agent was hanging because the driver threads were still running. I updated the CassandraNodeComponent class to implement the ResourceComponent.shutdown method where it shuts down the driver's thread pool. With this change, my agent exited gracefully. This change will be available in Jenkins build 225 and later.
Hi, I am not sure if the steps I took are not too rapid, but rhqctl stop either throws exception at the end of the process or is hanging while stopping agent. Please get attached log of all actions a logs taken. Moving back to ON_Dev
Created attachment 767692 [details] rhqctl_start_stop_hangs.png
Armine, I am not able to reproduce this issue. It may be due to the plugins (or rather the lack of plugins) that I am running. I typically test with a minimal set of plugins. When I previously commented on this issue, there cassandra plugin was not shutting down the datastax driver. That was a plugin-specific issue. There could be another plugin that is causing problems with the shutdown. If you see the issue again, can you provide the list of plugins you are running?
Hi John, I have installed rhq using rhqctl install, inventoried agent to server gui, called rhqctl stop --agent and it cannot shut down agent for ~15 mins. Please get attached screen-shot of rhqctl stop and fragments from server and agent logs.
Created attachment 774124 [details] rhqctl_stop_agent.png
Created attachment 774125 [details] rhqctl_stop_agent_agent.log
Created attachment 774126 [details] rhqctl_stop_agent_server.log
I am removing this from the Cassandra tracker and removing me as the assignee since the issue is not specific to rhqctl or the Cassandra feature work.
Armine, I'm closing the issue because we don't have enough elements to analyze it. If you manage to reproduce the problem, then please take a thread dump of the agent and reopen the BZ. Thanks