In the live session on my laptop I couldn't connect to the internet because I couldn't install the bcmwl-kernel-sources package. Apparently it tries to connect to the internet to install some of the packages. I'll try this with an ubuntu and kubuntu iso after and see if it is a generic issue or one that is specific to neon.
Description
Description
Related Objects
Related Objects
- Mentioned Here
- T8638: ISO build having dns problems
Comment Actions
Checked both Ubuntu and Kubuntu 18.04 and both setup the bcmwl-kernel-source correctly either by installing it directly with apt install bcmwl-kernel-source or the easier way of ubuntu-drivers autoinstall so there is definitely an issue on kde neon for it.
Comment Actions
May be T8638 as I now realize offline install works unless versions go out of sync -.-
What's the error?
What does apt policy bcmwl-kernel-source show?
What's in the /var/log/apt/* logs when you try to install it?
I am almost certain the problem is that the issue described in the other task results in the apt cache not updating and it therefore isn't using the on-ISO repo to get the deb from. Tricky to solve though. Also if that is the case it needs a more reliable addition to the offline qa test.
Comment Actions
If you do a sudo apt install bcmwl-kernel-sorce on a KVM install you'll see the issue. As explained though the issue is not present on kubuntu 18.04 so my assumption is the something like dkms is older on the ISO and not compatible with the driver so tries to connect to the online repo to get the correct version.
I'll try and get some logs tonight if I have time.
Comment Actions
Ok, I've managed to construct a test for this.
I don't think this ever reliably worked though, the actual logic pulling in packages is flawed it seems.
Comment Actions
Ah! oh! This is actually a regression because the tooling used to build the bionic iso is out of date 🙄
Comment Actions
If you can sort it today and spin out an image today I can grab it tonight and ensure it works.
Many thanks
Comment Actions
New ISO in the usual place. I can't say if it actually fixes the issue as there's been other regressions in the meantime preventing auto QA from passing, but the build log certainly looks more correct than before.