-
Bug
-
Resolution: Unresolved
-
Major
-
None
If the EC2 plugin runs into a "Failed to serialize" error when trying to provision a new node, it will leave that provisioning thread hanging and leaves a EC2 instance running(this may also be happening during other errors in provisioning).
The instance begins with:
`Looking for existing instances with describe-instance: $MY_TEMPLATE_STUFF`
`No existing instance found - created new instance: $MY_ACTUAL_TEMPLATE`
then that fails with:
SEVERE: Timer task hudson.slaves.NodeProvisioner$NodeProvisionerInvoker@719c5634 failed java.lang.RuntimeException: Failed to serialize hudson.plugins.ec2.EC2AbstractSlave#tags for class hudson.plugins.ec2.EC2OndemandSlave hudson.util.RobustReflectionConverter$2.writeField(RobustReflectionConverter.java:256)
Where it is failing to write a serializable field when saving the node.
What I would like to see is that if this node fails to provision/save then the EC2 plugin is able to clean up that created EC2 instance and remove the thread.
I think this is related to https://issues.jenkins-ci.org/browse/JENKINS-31804 which should probably be a EC2 plugin defect rather than a core one