Product Documentation

Troubleshooting

Sep 06, 2013

Before you attempt to resolve an AutoScale issue, make sure that the prerequisites have been adhered to, on both the CloudPlatform server and the NetScaler appliance, as described in "Prerequisites." If that does not resolve the issue, your problem could be one of the following.

The AutoScale configuration was successfully configured in CloudPlatform. Yet, the minimum number of VMs has not been created.

  • Recommend that the CloudPlatform user deploy one VM manually in the network before configuring AutoScale. Ask the user to remove the AutoScale configuration from the NetScaler appliance or the load balancer from the network, manually deploy one VM (preferably using the template created for the AutoScale configuration), and then create the AutoScale configuration.
  • Verify that the CloudPlatform user has configured the VM template in such a way that the VMs that are created from the template can accept traffic without manual intervention. If a provisioned VM cannot accept traffic automatically, the metric remains above the threshold, and the AutoScale configuration continues to provision additional VMs, as designed. To remedy the issue, disable AutoScale from CloudPlatform, fix the template, and then enable AutoScale.
  • Verify that the CloudPlatform user has not exceeded the limit for the number of VMs imposed by the user's account.
  • Verify that the CloudPlatform server is up and is reachable from the NetScaler appliance.
  • Verify that the CloudPlatform log file, management-server.log, has reported the successful creation of the AutoScale configuration in CloudPlatform.
  • Verify that the scale-up policy that is responsible for initial scale up (the policy name is prefixed with Cloud-AutoScale-Policy-Min) is receiving hits.

The AutoScale configuration is rapidly spawning a large number of VMs

  • Verify that the CloudPlatform user has configured the VM template in such a way that the VMs that are created from the template can accept traffic without manual intervention. If a provisioned VM cannot accept traffic automatically, the metric remains above the threshold, and the AutoScale configuration continues to provision additional VMs, as designed. To remedy the issue, disable AutoScale from CloudPlatform, fix the template, and then enable AutoScale.
  • Verify that the quiet time that the CloudPlatform user has configured in the AutoScale configuration is sufficient to ensure even traffic distribution to all the VMs, including the new VM. If the quiet time is too low, and traffic distribution has not stabilized, the metrics might remain above the threshold, and additional VMs might be spawned.

When I ran the top command on my VM, I noticed that the CPU usage on my VM had breached the threshold that was configured for the scale-up action in AutoScale. Yet, the application is not scaling up.

  • Verify that the CloudPlatform user has installed an SNMP agent in the VM template, and that the SNMP agent is up and running on every VM.
  • Verify that the CloudPlatform user has not exceeded the limit for the number of VMs imposed by the user's account.
  • Verify that the CloudPlatform user has correctly configured the SNMP parameters to collect metrics from the VM (for example, the community string and the port).
  • Verify that the scale-up or scale-down policy is receiving hits.
  • Verify that the CloudPlatform server is up, and that the CloudPlatform server is reachable from the NetScaler appliance.

One or more additional VMs have been created, but they are not accepting traffic (that is, VMs have been created, but the average value of the metrics is still above the threshold)

  • Verify that the user has configured the templates in such a way that the VMs created from the templates can start serving traffic without any manual intervention.
  • Verify that the service is running on the VMs, on the configured member port.
  • Send a ping request to the gateway (virtual router), from the VM that is not accepting traffic.

The AutoScale configuration has been deleted, but the VMs continue to exist

  • The VMs might not be deleted immediately after the AutoScale configuration is deleted. Wait for about 5 minutes after you have deleted the AutoScale configuration, and then check again.
  • If the destruction of VMs has not commenced after 5 minutes, you might have to delete the VMs manually.