'Is there a way to find out preemptible resource availability for different regions in Google Cloud?
I would like to utilize preemptible VM instances in Google Cloud, but sometimes run into supply issues, especially for GPUs
Is there any way to find out what data-center region usually has the best availability for certain (preemptible) resources?
As a even load on the data-centers should be in Google's interest I wonder why there is no such tool easily available. I could not find one at least.
Solution 1:[1]
To know the availability of GPUs in different regions please follow the doc.
Also sometimes the error that you are facing “supply issues” is due to the quota, because regarding the quota, it was set to 0 in most of the projects in GCP to avoid abuse of resources and you can request this for increase using the console. The error when deploying with GPU, means that the region you have selected has the resource for the machine type you've chosen and you only need to have a limit of 1 for GPUs in all regions to proceed. You can always request an additional quota limit for GPUS all regions.
GPUs attached to preemptible instances work like normal GPUs but persist only for the life of the instance. Consider requesting dedicated Preemptible GPU quota to use for GPUs on preemptible instances.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|---|
| Solution 1 | Bakul Mitra |