I am seeing that some times Jenkins jobs are stuck in queue for no reason. Per my observations Jenkins is not triggering new slave request for the label.

To be specific Jenkins not making a call to canProvision API ( http://javadoc.jenkins-ci.org/hudson/slaves/Cloud.html#canProvision(hudson.model.Label) ).

I tried below scripts to reset queue but not seeing any difference.
hudson.model.Hudson.getInstance().getQueue().scheduleMaintenance()
jenkins.model.Jenkins.getInstance().getQueue().scheduleMaintenance()

This is observed for both Amazon EC2 plugin and Azure slave plugin with Jenkins version 1.630.

Note that if i change the label in template then Jenkins is triggering new slave creation request. Even Jenkins restart is not helping.

To me it looks like some sort of deadlock condition on the label , what could be the possible reasons and is there any work around?

Any pointers on this will help me to debug the issue further