[Nix-dev] Debugging NixOS login problems

Rickard Nilsson rickard.nilsson at telia.com
Mon May 9 09:22:49 CEST 2011


 On Sun, 08 May 2011 22:49:57 +0200, Eelco Dolstra 
 <e.dolstra at tudelft.nl> wrote:
> Hi,
>
> On 05/06/2011 01:32 PM, Rickard Nilsson wrote:
>
>> I have a sort-of working NixOS installation booting from an NFS root
>> with only minor manual interventions. I pass init stage 1 and 2 just
>> fine, and then get a login prompt. Now if I try to login with root 
>> or
>> any other user, absolutely nothing happens. The login process just
>> hangs, and nothing happens if I press Ctrl-C. I can switch to 
>> another
>> tty and try to login there too, but nothing happens there either.
>
> Probably the system locked up because the dhclient Upstart job
> brought down the network interface used to access the NFS disk.  
> Since
> the code to subsequently bring the interface up again resides on NFS,
> it hangs. I've had this problem in our VM tests (where we boot from a
> remote Nix store mounted via SMB).  It uses this workaround (from
> modules/virtualisation/qemu-vm.nix):
>
>   # Starting DHCP brings down eth0, which kills the connection to the
>   # host filesystem and thus deadlocks the system.
>   networking.useDHCP = false;

 Thanks! I'll try it out when I get back home. Actually, I should have 
 thought about it myself, because I have a similar workaround on my 
 current ArchLinux setup, where I had to disable all networking 
 initialization on startup otherwise the link that was set up by initrd 
 would be teared down. Of course, I had forgot all about that by now.

   / Rickard



More information about the nix-dev mailing list