[Nix-dev] NixOS won't boot anymore in certain generations, don't know why (Stage-1 error)

[Matthias Beyer]

Hi,

I found and solved the issue.

So first, I checked the kernel versions in my revisions:

    106 - 3.18.11
    107 - 3.18.13
    108 - 3.18.13
    109 - 3.18.13
    110 - 3.18.13
    111 - 3.18.13
    112 - 3.18.13

(the latest two were other test builds).

I guessed that the issue was the kernel ... because nothing else
changed. There was no channel update and I didn't fiddle around in the
configuration for the system (at least not in these parts, only
packages and containers a bit).

So, I got my local clone of the nixpkgs repo onto the commit which was
used to build generation 106:

    2d8cfe76a9e4f05e391d30f1654d45dee5993b8a

And rebuild the system. This worked. Reboot worked too, so the new
revision, 113, is now on kernel 3.18.11 again.

I then tried to change kernel versions to 4.0 and 3.19 (in this order,
because newer is better), but both failed to build because the ati
driver can't be build because of _compiler errors_ (seriousely, what
the hell?)

So, I'm back on 3.18.11. Unfortunately, I have own patches because I
have to wait for the channel update. So I re-applied my patches upon
mentioned commit:

    gitolite v3.6.2 -> v3.6.3
    snort: version fix (fixes download error)
    daq: version fix (fixes download error)

these are non-published commits.

---

That's it. While I don't have any concerns about my machine, because
Nix is great and everything and I can boot in old versions of the
system and fix everything and so on, I really can't consider this as
great user experience. Debugging the issue was not that complicated
(despite my little hangover), but unneccessary. I'd really like to see
this situation improved. And I'd really like to see kernel 4.0 on my
machine, as I'm waiting for the AMD graphic cards fixes which are in
4.0.

I mean, I build my system from my own clone of nixpkgs. Not the way I
want to:

    nix-channel --update
    nixos-rebuild switch

which should be used. No, I have to

    git checkout <commit>
    nixos-rebuild switch -I nixpkgs=~/my/clone/of/nixpkgs
    # build fail
    # try again other commit

I don't want to blame anyone for this or something, this mail is
mainly for documenting the issue, but hey, these things really
shouldn't happen, right?

So, to close this here and now... I hope I see some of you guys next
weekend at tuebix in Tübingen, Germany!

On 04-06-2015 15:06:51, Matthias Beyer wrote:
> Hi,
> 
> I have a problem with some of my generations.
> 
> Today (, after installing chromium, but I don't think this has
> anything to do with it), I noticed that my xterms got redrawn after
> switching to them (I'm using i3, so if a window is not shown but I
> bring it up, a redraw happened).
> 
> So, if I switched to a xterm where I executed something like
> "alsamixer" or "tree /tmp", it got redrawn line by line.
> 
> I did not understand why this happened, but I guessed it was some
> driver problem, where a driver went bad or something. So I decided to
> reboot.
> 
> When rebooting, I booted into my newest generation, which was 109 by
> this time. But I got an error in stage 1, telling me that my root
> partition couldn't be mounted as the device did not come up (LUKS
> encrypted SSD, root on /dev/sda2). It asked me
> 
>     "dm_mod" loaded?
> 
> So I tried previous generations and had success with 106 (107-109) did
> not work. I checked my config, and I saw:
> 
>     boot.initrd.kernelModules = [ "fbcon" "ext4" "dm_crypt" ];
> 
> that "dm_mod" was missing, indeed. So I changed it to:
> 
>     boot.initrd.kernelModules = [ "fbcon" "ext4" "dm_mod" "dm_crypt" ];
> 
> And rebuild the system, resulting in generation 110. I tried to boot
> that, but the same error happened.
> 
> I'm on kernel 3_18_4, if this matters.
> 
> So my problem is, I don't know what went wrong and how to fix it.
> Unfortunately, I don't know which configuration I build generation 106
> from (my config is git-tracked). I'd show you the diff of my
> generation, but well... I don't know which revision it was.
> 
> How can I debug this and more important: How can I fix this?
> 
> -- 
> Mit freundlichen Grüßen,
> Kind regards,
> Matthias Beyer
> 
> Proudly sent with mutt.
> Happily signed with gnupg.



> _______________________________________________
> nix-dev mailing list
> nix-dev at lists.science.uu.nl
> http://lists.science.uu.nl/mailman/listinfo/nix-dev


-- 
Mit freundlichen Grüßen,
Kind regards,
Matthias Beyer

Proudly sent with mutt.
Happily signed with gnupg.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
Url : http://lists.science.uu.nl/pipermail/nix-dev/attachments/20150604/c00dfcb4/attachment.bin