Daniel Golle

@hnyman This looks like https://github.com/openwrt/openwrt/commit/72accd078f has not yet been applied to the container used for CI. How can we know which commit level the SDK used in the container is...

The main difference is that the LuCI interface uses your browser to communicate with the asu server -- in this way you can update even devices which are not directly...

> Do we really need "correct" macaddr in failsafe mode? It's better imho as it makes the device identifiable on the network.

> random mac address shouldn't be a problem. The random mac address *is* a problem if you rely on multicast ping replies to identify the device inside a network shared...

> mhhh why the opposite doesn't work? Because the initramfs image is not generated individually for each device even if `CONFIG_TARGET_PER_DEVICE_ROOTFS` is set. It's only generated once for the whole...

> btw there was this old project that was never finished that just address the problem Oh nice, I was completely unaware of that. Doing this would of course be...

> Think it has been abandoned but I can take over If you want to do that it would be great, of course. Otherwise I will go head and merge...

Closing in favor of https://github.com/openwrt/openwrt/pull/12959

The problem you describe does not seem to happen on the buildbots. Also, dependencies are selected automatically and do not need to be explicitly listed. Also note that `CONFIG_THERMAL=y` for...

What ever the problem may be, adding kmod-thermal as module for the BPi-R4 is definitely not the solution. Edit: Likely the solution would be to follow-up a newly added hidden...