Ambari stuck with "Install Pending" when creating cluster thru api

[#1]

version: Ambari 1.7.0

Have anyone seen error below when trying to deploy a new cluster using ambari api.

30 Nov 2015 17:24:06,826  WARN [ambari-action-scheduler] ActionScheduler:200 - Exception received
java.lang.RuntimeException: org.apache.ambari.server.ClusterNotFoundException: Cluster not found, clusterName=clusterID=6
        at org.apache.ambari.server.actionmanager.ExecutionCommandWrapper.getExecutionCommand(ExecutionCommandWrapper.java:116)
        at org.apache.ambari.server.actionmanager.ActionScheduler.isStageHasBackgroundCommandsOnly(ActionScheduler.java:428)
        at org.apache.ambari.server.actionmanager.ActionScheduler.filterParallelPerHostStages(ActionScheduler.java:416)
        at org.apache.ambari.server.actionmanager.ActionScheduler.doWork(ActionScheduler.java:246)
        at org.apache.ambari.server.actionmanager.ActionScheduler.run(ActionScheduler.java:195)
        at java.lang.Thread.run(Thread.java:745)
Caused by: org.apache.ambari.server.ClusterNotFoundException: Cluster not found, clusterName=clusterID=6
        at org.apache.ambari.server.state.cluster.ClustersImpl.getClusterById(ClustersImpl.java:250)
        at org.apache.ambari.server.actionmanager.ExecutionCommandWrapper.getExecutionCommand(ExecutionCommandWrapper.java:74)
        ... 5 more
30 Nov 2015 17:24:09,931  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0233pl.redecorp.br
30 Nov 2015 17:24:09,931  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0234pl.redecorp.br
30 Nov 2015 17:24:09,932  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0232pl.redecorp.br
30 Nov 2015 17:24:09,932  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0237pl.redecorp.br
30 Nov 2015 17:24:09,932  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0235pl.redecorp.br
30 Nov 2015 17:24:09,932  WARN [ambari-hearbeat-monitor] HeartbeatMonitor:146 - Heartbeat lost from host brtlvlts0236pl.redecorp.br
30 Nov 2015 17:24:13,274 ERROR [alert-event-bus-1] AmbariJpaLocalTxnInterceptor:114 - [DETAILED ERROR] Rollback reason:

If we log in ambari webui, cluster stuck with install pending:

[#2]

@Guilherme Braccialli

1. java.lang.RuntimeException: org.apache.ambari.server.ClusterNotFoundException:Clusternot found, clusterName=clusterID=6

check if agents are running. Verify the commands again.

[#3]

With Ambari 1.7.0, all hosts must be registered prior to submitting a blueprint cluster creation request. If the necessary hosts aren't available in 1.7.0, the request will immediately fail and indicate to the user that the necessary hosts aren't available. In more recent Ambari versions, it is possible to submit a cluster create request prior to any hosts being available and until "the minimal required hosts" become available, you are correct that the request state will remain "PENDING".

[#4]

Can you share the commands that you used to setup the cluster?

[#5]

@Jonas Straub I'll try to get the commands, it's from Vmware Big Data Extension.

[#6]

Can you please confirm that you are deploying the cluster using the blueprint API and not the lower level 'primitive' API's? I see that this is for a rather old version "1.7.0", which was one of the first versions to support blueprint installs. I don't remember ever seeing this error with 1.7.0 BP installs so you will need to attach the ambari server log as well as confirm the API calls that resulted in this issue.

[#7]

@Guilherme Braccialli are you still having issues with this? Please close.

[#8 · Accepted solution]

@Artem Ervits

we moved to ambari 2.1.2 and issue never happened again.

[#9]

@jspeidel

so if we want to install few additional services ABC after a cluster is setup with few XYZ services , it is not possible?? I am unsure of the technicalities but this way it will be a problem for companies running analytics as a service

[#10]

I tried restarting (stop & start) all services in the cluster , during that process, the pending installation kicked off and wassuccessful I am not sure of the internal architecture