Akila created STRATOS-1489:
------------------------------
Summary: Faulty behavior in application monitor creation logic
Key: STRATOS-1489
URL: https://issues.apache.org/jira/browse/STRATOS-1489
Project: Stratos
Issue Type: Bug
Affects Versions: 4.1.0
Reporter: Akila
When a user sign up for an application, AS makes a service call to CC and
executed "createApplicationClusters" method which will publish
ApplicationClustersCreated event. There is a separate event listener for this
event in AS side which will start cluster monitors upon receiving this event.
But in this listener AS first checks whether this corresponding application's
status is set to DEPLOYED state. If not it will wait for 5 seconds and print
following warning;
WARN
{org.apache.stratos.autoscaler.event.receiver.topology.AutoscalerTopologyEventReceiver}
- Application monitor creation is not triggered on application clusters
created event even after 5 retries [application-id] xxxxxx-xxxxx. Possible
cause is either application context is null or application status didn't become
Deployed yet.
In the code [1] there is no way to recover if the application doesn't get
updated to DEPLOYED status within the given 5 seconds timeout window. There is
a comment in the code which says that the reason for not throwing an error is
because this is not the only place where it starts monitors. But the only other
place I could find is upon receiving CompleteTopologyEvent. But that event is
processed only at the very first time.
Therefore, there is no way to recover from this error state, and it does not
throw an error either. This logic needs to be improved.
[1]
https://github.com/apache/stratos/blob/master/components/org.apache.stratos.autoscaler/src/main/java/org/apache/stratos/autoscaler/event/receiver/topology/AutoscalerTopologyEventReceiver.java#L169
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
