1 |
On Thu, Aug 12, 2010 at 3:40 PM, Mike Frysinger wrote: |
2 |
> On Thu, Aug 12, 2010 at 3:33 PM, Eray Aslan wrote: |
3 |
>> It will be somewhat more work but instead of the above, we can say "tmpfs |
4 |
>> might be used for /var/run and /var/lock and the init scripts should handle |
5 |
>> this correctly". It feels (for want of a better word) better. |
6 |
> |
7 |
> i certainly use a tmpfs on /var/run to minimize disk writes. packages |
8 |
> that break i file bugs for and/or fix myself. it isnt that hard. |
9 |
> |
10 |
> plus, it's just good behavior. if /var/run gets removed for whatever |
11 |
> reason, people have to re-emerge a bunch of packages to simply create |
12 |
> a subdir ? that's silly. |
13 |
|
14 |
for people who do wish to improve their init.d scripts, recreating a |
15 |
dir is easy if your init.d runs after localmount: |
16 |
[ ! -d /var/run/foo ] && install -d -m 755 -o fowner -g fgroup /var/run/foo |
17 |
|
18 |
if your init.d runs before localmount, you'll have to resort to normal |
19 |
mkdir/chown/chmod, but i dont think there are many (any?) scripts |
20 |
that'll hit this set of requirements |
21 |
-mike |