[aws provider] Investigate why 5000 instances were created during the load testing
Categories
(Taskcluster :: Services, task)
Tracking
(Not tracked)
People
(Reporter: owlish, Assigned: bstack)
References
Details
I repeated the 1000 instances test with t2.nano instances. AWS provider spun up around 5000 instances
My initial assumption was it happened because the registration takes long, but it's not seconds-fast even in the case of one instance. However, it doesn't create additional instances for workerPools with 1 or 10 instances. Let's make sure of the reason it happened
|
Reporter | |
Updated•5 years ago
|
|
|
Reporter | |
Comment 1•5 years ago
|
||
The workerscanner fails to update the amount of running instances correctly. In the logs, I see seen: 44 and then lots of RequestLimitExceeded errors. It isn't an issue for 100 instances, so apparently 1000 is just a bit too much
|
|
Reporter | |
Updated•5 years ago
|
|
|
Reporter | |
Comment 2•5 years ago
|
||
Assigned to bstack as he's now working on changing how the estimator works
|
Assignee | |
Comment 3•5 years ago
|
||
I believe this is solved for now. If we see things like this again we can take some more radical approaches to keeping everything in agreement.
Description
•