Any chance you could get on the instance after that happens and see what shows up in /var/log/secure, say?

Sure enough! I found some logs in /var/log/auth.log mentioning attempted logins to root. Couldn't figure how how to have the plug-in log in via the 'ubuntu' user but still use the public key authentication. But by setting disable_root to 0 in /etc/cloud/cloud.cfg in my amazon image, I seem to be past that now I have a simple job running, but still having some problems. When the first job is queued, it starts a new amazon machine to service it. However if I put a second job in the queue, when it tries to start the second amazon machine I get this stack trace:

[#|2012-11-06T11:48:08.801-0500|WARNING|oracle-glassfish3.1.1|hudson.slaves.NodeProvisioner|_ThreadID=26;_ThreadName=Thread-2;|Provisioned slave testaws failed to launch
java.lang.NullPointerException
at java.util.TreeMap.put(TreeMap.java:541)
at java.util.TreeSet.add(TreeSet.java:238)
at hudson.model.Node.getAssignedLabels(Node.java:240)
at hudson.model.Slave.<init>(Slave.java:154)
at hudson.model.Slave.<init>(Slave.java:132)
at jenkins.plugins.jclouds.compute.JCloudsSlave.<init>(JCloudsSlave.java:47)
at jenkins.plugins.jclouds.compute.JCloudsSlave.<init>(JCloudsSlave.java:70)
at jenkins.plugins.jclouds.compute.JCloudsSlaveTemplate.provisionSlave(JCloudsSlaveTemplate.java:163)
at jenkins.plugins.jclouds.compute.JCloudsCloud$2.call(JCloudsCloud.java:216)
at jenkins.plugins.jclouds.compute.JCloudsCloud$2.call(JCloudsCloud.java:213)
at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
at java.util.concurrent.FutureTask.run(FutureTask.java:138)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)

Any chance you could attach the jclouds section of your config.xml?

jclouds section of config.xml, with private key and authentication stuff removed.

If you give me your amazon user number, I can share the AMI with you.

Is that the account number?

I think so? I've never shared an AMI before. The field I have to enter is labeled "AWS Account Number". Thanks for all your help!

0041-2815-1137, then. =) I'll try to give this a spin later today/early tomorrow - I couldn't see anything obviously wrong in the config, and it's probably not the specific AMI, but it's worth a try.

Thanks, you should have permission to access the AMI now.

I just tried it again to make sure I wasn't crazy. This time I don't get any exceptions. According to the log it starts up 3 slaves on the cloud, but they don't appear in the GUI, and the tied job is still hanging. How is it supposed to differentiate different instances of the same "testaws" node?

Again, thanks so much for your help.

Out of date.
If the problem persists with the upcoming v2.9, please open a new issue.