Closed Bug 1466202 Opened 7 years ago Closed 7 years ago

AWS Provisioner takes several minutes to provision instances after worker type changes

Categories

(Taskcluster :: Services, defect)

Product:
Taskcluster
Component:
Services
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
RESOLVED FIXED

People

(Reporter: gps, Assigned: jhford)

Details

Last night, I changed the gecko-2-decision, gecko-3-decision, and gecko-1-images worker types to use c5d instances (see bug 1462528). I made the change around 0015 UTC. It took ~45 minutes for new instances to spawn. Why, I'm not sure. Today, I changed the gecko-1-images worker type again to take into account my newly learned knowledge from bug 1466186. Again, it took 10+ minutes for a new instance to be spawned. This isn't the first time I've seen the AWS provisioner seem to react extremely slowly to worker type changes. I'm not sure if this is a feature or a bug or what reasonable expectations should be. But ~45 minutes seems quite excessive. So I'm filing a bug for the issue.
Greg, as long as existing instances are alive with the old configuration, the new one will not start. This is intentional, the provisioner only knows how to create instances, worker are responsible for ending themselves. What should help here is using the terminate all instances button in the UI.
Status: NEW → RESOLVED
Closed: 7 years ago
Resolution: --- → FIXED
Assignee: nobody → jhford
Component: AWS-Provisioner → Services
You need to log in before you can comment on or make changes to this bug.