lemaral

Thanks! Following your guidance, I can do the PR if it's ok

@shivamerla thank you for your reply, yes it never stops, although the gpu operator itself is working perfectly. I am seeing this using Falco and had to disable the relevant...

I would like also to use different channels for separate namespaces, to allow developers to monitor only their related workloads instead of having to watch everything

Is hostNetwork necessary then? It seems to work without

@linki, thanks for the follow-up, indeed A/CNAME/ALIAS and the corresponding TXT record must stay in the same batch to keep operations atomic. My concern (as you highlighted it) is definitely...

@szuecs, indeed binary search seems the most efficient wrt api rate limiting, we only need to ensure we're not looping forever. For the question you brought up, can you elaborate...

> Also `podman machine start` apparently respects XDG_DATA_HOME and puts the API forwarding socket in the right location but podman desktop looks for the API socket in this location `~/.local/share/containers/podman/machine/qemu/podman.sock`...