Actual availability of instance type in region

0

Hello all, I am currently running SageMaker for training jobs. Recently, when trying to launch new training job instance (specifically g5.12xlarge) I receive: CapacityError: Unable to provision requested ML compute capacity. To mitigate this I am thinking of switching to a different region. Before I do so, I would like to check whether I will be better off in the new selected region

My question is: other than trial & error, is there a way to understand the actual availability (taking into account the number of instances and usage) of a desired instance in a specific region? Thank you

1 Risposta
0

There's no document that shows actual instance availability for region as it's a pretty dynamic number. SageMaker is backed by EC2 instances for training jobs and when you receive that error it just means there isn't currently existing EC2 capacity for your specific use case. Here's a document on troubleshooting that specific error. You'll notice that the first troubleshooting is just waiting a few minutes then try again. That's generally because we don't always have large instance types just running idle, so once you receive that error, if you wait a few minutes, AWS works on spinning up capacity to meet your need. Sorry for the inconvenience, but I'd try those steps before moving region, as you may experience the same error in other regions as well.

AWS
AWSJoe
con risposta 6 mesi fa
profile picture
ESPERTO
verificato un mese fa

Accesso non effettuato. Accedi per postare una risposta.

Una buona risposta soddisfa chiaramente la domanda, fornisce un feedback costruttivo e incoraggia la crescita professionale del richiedente.

Linee guida per rispondere alle domande