You can find below the most common issues (and their solutions) you may face when deploy entities provided by Clocker. For more general troubleshooting, please check out the general troubleshooting or AMP docs.

Persisted State

If you have previously run AMP or AMP you may have to remove the persisted state these have saved on your machine. To do this, delete the folder ~/.brooklyn/brooklyn-persisted-state/ and all of it’s contents.

AWS VPC / Classic

It’s recommended that you use VPC mode to run Clocker within AWS. This is because problems have been recorded with DNS and starting instances withn AWS Classic. For more information on doing this, see the AMP docs.

AWS 401 errors

If you are getting 401 response errors when trying to provision machines make sure that: * The credentials you are using have the permission to create instances * You have accepted the TOC for the image you are using in the location definition. Go to and proceed to launch an instance with that image. In the process you’ll be asked to accept the TOC. You can abort the launch at this point.

High CPU usage

After a few hours of management of a swarm or cluster, AMP or AMP may start to suffer from high CPU and memory usage. This is because of a known issue in the Java garbage collection system. To fix this add the following to your AMP or AMP:

Open bin/amp and change the line

JAVA_OPTS="-Xms256m -Xmx2g -XX:MaxPermSize=256m"

Open bin/setenv and edit or add the line to the start

export JAVA_MAX_MEM="2G"

Failed to find machine-unique group on node

This is caused because of a restrictive security group within your cloud. Add a new piece of config to your swarm or cluster in the form kubernetes.sharedsecuritygroup.create or swarm.sharedsecuritygroup.create and set this to false.

Then create a new security group in your cloud and specify this in your location using the securityGroups key.

For more information on setting up a location, see the Getting Started guide.