| 1 |
Michael Hampicke wrote: |
| 2 |
> Am 09.09.2013 21:05, schrieb Benjamin Block: |
| 3 |
>> On 08:30 Mon 09 Sep , Michael Hampicke wrote: |
| 4 |
>>> Am 08.09.2013 20:51, schrieb Benjamin Block: |
| 5 |
>>>> Hej folks, |
| 6 |
>>>> |
| 7 |
>>>> I wonder what is a good way to create an image of a gentoo-system, so |
| 8 |
>>>> that one can apply it later to the same or other computers. |
| 9 |
>>>> |
| 10 |
>>>> In my case it is a rather simple setup: one partition, no encryption or |
| 11 |
>>>> lvm. Its a debug-setup, so its only used for certain programming-tasks |
| 12 |
>>>> and not for daily work, so no need for something fancy. The time I |
| 13 |
setup |
| 14 |
>>>> that system I also used only conservative compilation-flags and |
| 15 |
>>>> optimisation, so that it can be used on other CPUs (well, they have to |
| 16 |
>>>> be x86_64 and have to have mmx/sse[23] - but I think every setup that I |
| 17 |
>>>> intend to use this on will have these properties). |
| 18 |
>>>> |
| 19 |
>>>> So I reckon that one could just use tar with |
| 20 |
permission-preservation and |
| 21 |
>>>> some excludes like dev/sys/proc/tmp. But is this a good idea or is |
| 22 |
there |
| 23 |
>>>> a better way to do this? I never cloned a gentoo-system, so thats why I |
| 24 |
>>>> would like to be at least somewhat sure about it, so that I don't have |
| 25 |
>>>> to reconfigure it later again, because I messed it up :D |
| 26 |
>>>> |
| 27 |
>>> |
| 28 |
>>> Tar with permission preservation is fine. Just exlude everything in |
| 29 |
>>> dev/sys/proc/tmp as you said. But make sure, that these directories are |
| 30 |
>>> in your tar file, it does not matter if they are empty, but they have to |
| 31 |
>>> exist in order to boot proplery. |
| 32 |
>>> |
| 33 |
>>> One special case. To boot you most likely will need /dev/console and |
| 34 |
>>> /dev/null. Just inlcude those two device nodes in your tar file. |
| 35 |
>>> |
| 36 |
>> |
| 37 |
>> Thanks for pointing that out, but why are these both special? Seems to |
| 38 |
>> me like these are also (char)device-nodes and shouldn't they also be |
| 39 |
>> generated by the kernel with DEVTMPFS and then udev at a very early |
| 40 |
>> init-stage? |
| 41 |
> |
| 42 |
> If you have DEVTMPFS enabled you should be fine. But not everybody has |
| 43 |
> that enabled, or even uses udev :-) |
| 44 |
> |
| 45 |
|
| 46 |
I would include them just in case. Why take the chance that it fails |
| 47 |
for whatever reason. |
| 48 |
|
| 49 |
Dale |
| 50 |
|
| 51 |
:-) :-) |
| 52 |
|
| 53 |
-- |
| 54 |
I am only responsible for what I said ... Not for what you understood or |
| 55 |
how you interpreted my words! |
Archives