Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is.
It appears possible in certain circumstances a vllm deployment can fail to be deleted - e.g. connection with cluster is interrupted at delete time. This can lead to exhausting available environments and everything halting,
Describe the solution you'd like
A clear and concise description of what you want to happen.
It might be useful to record delete requests so the env manager can periodically check if a deployment that was supposed to be deleted is still present and garbage collect it.