[Nix-dev] NixOS hangs on boot after upgrading 15.09 -> 16.03; systemd update to blame?

[James Cook]

The discussion has moved to the GitHub issue; we've made significant
progress. I will run the script that regenerates
hardware-configuration.nix at some point though.

On 25 June 2016 at 02:18, zimbatm <zimbatm at zimbatm.com> wrote:
> Try re-running nixos-install to regenerate the hardware.nix file. If that
> doesn't work I guess it depends on your config and hardware so more info on
> that would be useful.
>
>
> On Thu, 23 Jun 2016, 11:55 Tomasz Czyż, <tomasz.czyz at gmail.com> wrote:
>>
>> James, maybe it's worth to open the ticket on
>> https://github.com/NixOS/nixpkgs/issues/
>>
>> 2016-06-23 4:08 GMT+01:00 James Cook <james.cook at utoronto.ca>:
>>>
>>> On 22 June 2016 at 19:27, James Cook <james.cook at utoronto.ca> wrote:
>>> > After updating my desktop from 15.09 to 16.03, it hangs on boot after
>>> > printing the message "regenerating udev hardware database". I do not
>>> > think the problem is directly related to that message; see below. Any
>>> > ideas for debugging this?
>>> >
>>> > Here's what I've figured out so far:
>>> >
>>> > * Using git-bisect, it seems that the problem starts at commit
>>> > ecbe04b4, "systemd: Update to 225". (I had to cherry-pick 3cc04956,
>>> > "libseccomp: Fix hash", to got it to build.)
>>> >
>>> > * I don't think the script that prints "regenerating udev hardware
>>> > database" is the one that hangs. I added a debugging echo command to
>>> > the end of system.activationScripts.udevd (in
>>> > nixos/modules/services/hardware/udev.nix), and I saw the extra message
>>> > when I booted. So the hang happens after.
>>> >
>>> > My next step was going to be to add more debugging statements to the
>>> > boot process, but I'm having trouble figuring out where any of that
>>> > stuff is. I can't find where system.activationScripts.udevd is called
>>> > from, and I was mystified to find that grep -R "egenerating udev"
>>> > /nix/store/(...)-nixos-16.0.3.git.3cc0495M produces no output (except
>>> > some errors, e.g. etc/zoneinfo/posix is a directoly loop).
>>> >
>>> > James
>>>
>>> Another note: the problem also happens on master (commit 0167eef9).
>>> The "regenerating udev hardware database..." message is no longer
>>> printed at master; the last messages before the hang are "setting up
>>> /etc..." and a bunch of "removing obsolete symblink ..." and "removing
>>> group ..." messages.
>>>
>>> James
>>> _______________________________________________
>>> nix-dev mailing list
>>> nix-dev at lists.science.uu.nl
>>> http://lists.science.uu.nl/mailman/listinfo/nix-dev
>>
>>
>>
>>
>> --
>> Tomasz Czyż
>> _______________________________________________
>> nix-dev mailing list
>> nix-dev at lists.science.uu.nl
>> http://lists.science.uu.nl/mailman/listinfo/nix-dev