Christophe de Dinechin

@lifupan I think you are onto something. I will try that on Wednesday.

I've tried yesterday, but for some reason this causes CRIO to crash now. ``` Nov 13 12:05:09 turbo-fedora systemd-coredump[376661]: [🡕] Process 1664862 (crio) of user 0 dumped core. Module /usr/bin/crio...

Hmm, I now see something that I don't remember seeing before, but that may be the root cause: ``` level=error msg="qemu-system-x86_64: -netdev tap,id=network-0,vhost=on,vhostfds=5,fds=6: Unable to query TUNGETIFF on FD 6:...

The reason I need to restart CRIO is itself a bit annoying: if I don't, the bad intermediate status of the pod leaves the name hanging around, apparently. I see...

> Hmm, I now see something that I don't remember seeing before, but that may be the root cause: > > ``` > level=error msg="qemu-system-x86_64: -netdev tap,id=network-0,vhost=on,vhostfds=5,fds=6: Unable to query...

@lifupan I've retried on and off, but for now I don't think it works, though I cannot reproduce the exact same symptoms as before, so there may be a bit...

OK, so after restoring my network configuration to the `tcfilter` situation where we had `Link not found` (I also tried `macvtap` to see if I could workaround), I now see...

I wonder if we should not also provide some reference networking configuration for CRI-O and containerd when running them from the command line.

I compared the qemu command-line in a cluster where things work and on a VM where things don't, and I don't see any difference that would explain why one works...

Relevant note: Chromium is not a possible choice anymore, it does not pass the Mac Gatekeeper tests, so you cannot launch it.