change the GPU pod type without recreating
Is there an option available whereby, if the previous GPU becomes unavailable, I can select another GPU type without recreating it? I'm confused because I need to ensure that my pod ID remains fixed.
2 Replies
I have the same question. Do I have to completely recreate a new instance and start over?
Wrote it in general, but will answer here:
1) If you stop your Pod, you release it back to the public the GPU. It is extremely competitive, so you will always have better luck trying to launch a new one. I personally use a pod + third party storage to make this easy. You can use network storage, but it locks you to a region and you can run into availiability problems.
2) Depending on your use case, on why you need the Pod ID to stay static, I guess, why are you releasing it in the first place / what is your use case?
Depending from there, you can either do serverless (a way to host an endpoint to process incoming requests on runpod dynamically), or set up a system to dynamically scale up and down a pod yourself, and grab the new pod id yourself.