[AMBARI-5036] Secured: Start All Services task got stuck forever

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.5.0
• Fix Version/s: 1.5.1
• Component/s: ambari-server, test
• Assignee: Dmitry Lysnichenko
• Reporter: Dmitry Lysnichenko
• Created: 11/Mar/14 20:33
• Updated: 21/Mar/14 14:14
• Resolved: 21/Mar/14 14:14

Description

1. Deployed 2-node cluster.
2. Added 3rd node.
3. Enabled security.

After steps above on all 3 hosts tasks jammed and don't want to perform or fail for a very long time.
VMs are alive, ambari-server and all ambari-agents are running.

Finally got a reproduce using 2 commands

curl 'http://vm-0.vm:8080/api/v1/clusters/cc/services?params/run_smoke_test=false' -X PUT -H 'X-Requested-By: X-Requested-By' -u admin:admin --data '{"RequestInfo": {"context": "Start All Services"}, "Body": {"ServiceInfo": {"state": "STARTED"}}}' ; sleep 3;  curl 'http://vm-0.vm:8080/api/v1/clusters/cc/hosts/vm-0.vm/host_components/APP_TIMELINE_SERVER' -X DELETE -H 'X-Requested-By: X-Requested-By' -u admin:admin

The way to reproduce is a bit different compared to an original description (I issue a DELETE request in 3 seconds after START_ALL_SERVICES request has been issued), but the symptoms are the same: ServiceComponentHostNotFoundException exception is posted to log and operation is stuck on stage that contains "App Timeline Server Start" command.

Issue Links

• links to: Review board