I've got an AMI init script in my ec2 slave creation configuration that takes several minutes to run - it copies down some large files from S3 and unpacks them in order to prepopulate hudson's workspace with checkouts to reduce scm server load.

Hudson is unhappy about this taking so long - it tries to start up another node before the first one is finished being set up; presumably it thinks it's dead or slow or something.

Suggestions for how to fix:

1. Is it possible to move the 'are you alive' check before the initscript? I guess it would have to be separate from the 'are you ready' check in that case, though.

2. Allow the user to specify the timeout

3. Do like hudson does for builds: instrument the first execution of the initscript(s) and set the timeout to some percentage (150%? 200%?) of that.

Using Hudson v1.339 and EC2 plugin v1.5.