Description of problem: When you start a stock EAP6.1.DR4 in domain mode, you cannot restart the servers of the main servers group. server-one restarts but server-two does nothing and the http request to the management interface blocks. Version-Release number of selected component (if applicable): EAP6.1.DR4 How reproducible: Always Steps to Reproduce: 1. Download and extract EAP6.1.DR4 2. Start in domain mode 3. Execute HTTP request with curl curl -v --data "{\"operation\":\"restart-servers\",\"address\":[{\"server-group\":\"main-server-group\"}]}" -H "Content-Type: application/json" -H "Accept: application/json" --anyauth -u rhq:rhqadmin_1 http://localhost:9990/management Actual results: server-one restarts, server-two does nothing, curl command blocks Expected results: server-one restarts, server-two restarts, curl command ends with operation result success. Additional info: If you try to restart the group through the admin console it also fails.
Brian Stansberry <brian.stansberry> made a comment on jira AS7-6711 It's simpler than the original description. Multiple servers aren't required, nor is restart-servers required. Calling /server-group=x:start-servers when the master HC has just a single server in group "x" will result in the problem. The problem is that: 1) OperationCoordinatorStepHandler calls context.acquireControllerLock() at line 209. The lock will be held until the server connects. 2) The registering server needs to acquire the lock as part of registration, ServerToHostProtocolHandler line 254. Until it gets the lock, it cannot connect. This deadlocks. A likely workaround is to not run servers on the master HostController. The lock acquired in 1) is done on the master.
Emanuel Muckenhuber <emuckenh> updated the status of jira AS7-6711 to Resolved
Verified in EAP 6.1.0 ER4