Ticket #1001 (new defect)
start_instance on a latent BuildSlave called before stop_instance finished
|Reported by:||moschny||Owned by:|
We are using a descendant of an AbstractLatentBuildSlave, which is basically modeled after LibVirtSlave.
The idea is to have it started if needed, execute exactly one build, then insubstantiate again.
Therefore, max_builds is set to 1, and the slave calls self.insubstantiate() in its buildFinished method.
However, there is a race (imho a bug): When multiple builds are to be performed, the slave's start_instance is sometimes called before stop_instance is finished (i.e. before its last deferred has fired).
- Keywords virtualization added
- Milestone changed from undecided to 0.8.3