[Nix-dev] Disaster recovery / re-deployment

[Rickard Nilsson]

 Hi,

 On Tue, 03 Jan 2012 15:47:23 +0100, Florian Friesdorf 
 <flo at chaoflow.net> wrote:
> Hi Rickard,
>
> On Tue, 04 Oct 2011 11:45:14 +0200, "Rickard Nilsson"
> <rickard.nilsson at telia.com> wrote:
>> > However in your case
>> > http://mawercer.de/~marc/minimal-install-archive.patch should be 
>> enough
>> > if you're lucky.
>>
>> Thanks! I will start out from that and see where I end up.
>
> did you made progress with that? For a full backup there might be 
> more
> files needed, e.g.:
> /etc/wpa_supplicant.conf
> /etc/passwd
> /etc/shadow
> /etc/group
> /var/lib/wicd/*

 I used Marc's bootstrapper and it worked, but I haven't yet got to that 
 perfect re-deployment system I want. I'm still working on basic stuff I 
 want to get in place before switching over to NixOS.

 However, thanks for the hints about various pieces of state spread out 
 all over the place, even in a Nix system :) On my current NixOS test 
 setups I actually run with the root filesystem in memory (tmpfs), and 
 then I am forced to solve these things as I discover them. So 
 essentially, I'm doing re-deployment on each reboot now, which is 
 exactly what I want. I haven't got the Nix store/db/profiles sharing 
 sorted out completely yet, but I think I'm getting there. My plan is to 
 end up with one master host that is the only one that can write to the 
 shared store and db. This host will then be the single point of backup. 
 All other hosts and workstations will boot from the network and mount 
 the store and db read-only over NFS or CIFS.


 Regards,
   Rickard