Closed ericjeske closed 1 year ago
More context:
I think the issue I'm running into with cluster creation failures (both for high worker count fargate clusters as well as ~medium count fargate_spot clusters) is a race condition with waiting for workers. I've found that if I create a small cluster (5 workers // 10 cores) and then scale them manually by 10-20 workers, I don't run into any issues
Fargate spot deprecated
Intermittent failures to create spot clusters (within owned AWS account).
The cluster was created in the coiled UI and workers appeared in ECS.
Attempted to spin up non-spot cluster without any issues. Subsequent (and intermittent) attempts to spin up a spot cluster were successful.