0

I subscribed to GCP and received the $300 credits. Then I upgraded my account to "paid account". Next, I increased the limit for multiple VM types with GPU, in multiple regions, and received approval emails.

I tried creating a VM with GPU in different regions, in all zones, with different configurations (T4, P4, etc), but all what I am getting is Operation type [insert] failed with message "The zone 'projects/xxxxxx/zones/us-central1-a' does not have enough resources available to fulfill the request. Try a different zone, or try again later."
That's only one example. As mentioned, I tried multiple regions/zones/config combinations, but no luck. Am I doing anything wrong???

cloud_IaaS
  • 1

1 Answers1

0

You're not alone, we've been trying to start an existing GPU instance since the 11th of December. Regardless of region or zone, instances with a GPU are not able to start.

It's absolutely unbelievable that Google would allow this to happen. The sole purpose of using a public cloud is to protect against infrastructure and capacity related issues. A temporary lack of capacity is acceptable but this has spanned weeks and doesn't seem to be isolated to a particular region.

The responses we've had from Google's support have been diabolical and outright lies at times. We were even told they had no persistent SSD storage available in eu-west4. How can this be allowed to happen? Google's capacity planning is shocking and demonstrates once again that public clouds are not what they're advertised to be.

Ryan
  • 1