Alpine image susceptible to connection timeout

[mnlipp]

When using the alpine based image for the runner, spice connections sometimes terminate with
Spice: rcc has been unresponsive for more than 30000 ms, disconnecting.

This happens usually during boot or when the screen is locked. It doesn't seem to be an issue while using the machine.

The problem has been mentioned in various reports (e.g. here) in different contexts on the internet without any resolution.

It's rather strange, because it happens only with the alpine image/package that uses exactly the same qemu version (8.0.4) as the arch image/package. Arch doesn't patch the original sources at all, and the patches from alpine seem unrelated to the issue. Most likely, the problem depends on some network issue of the underlying OS.

The default image for the runner has been set to the arch image, despite its huge size. I recommend to test the alpine image nevertheless, maybe it's possible to narrow things down.