[HEADS-UP] systemd for F14 - the next steps

[Daniel J Walsh]

On 07/14/2010 03:20 PM, Lennart Poettering wrote:
> On Wed, 14.07.10 14:24, Daniel J Walsh (dwalsh at redhat.com) wrote:
> 
>>>> myapp_t creating a directory in var_run_t will be labeled
>>>> myapp_var_run_t.  I would just need to go through all the policy that
>>>> uses var_run_t directories and make sure it has this rule.
>>>
>>> Hmm, if you would be willing to do that, then it would be great to find
>>> somebody who fixed the .specs and makes a list of packages whose selinux
>>> policy needs fixing. Anyone? Rahul you should vague interest on IRC?
>>>
>>> Lennart
>>>
>> What is the big benefit of changing to tmpfs for /var/run?
> 
> Well, various little things:
> 
> The reboot cleanup of those dirs will become unnecessary and automatic.
> 
> Socket accesses won't put pressure on the HDD due to atime updates.
> 
> Security reg. left-over files
> 
> Stateless bootup
> 
> Multi-instance root fs with r/o mounts.
> 
> And stuff like this. And of course this is just cleaner this way, since
> the files in /var/run and /var/lock are runtime objects that are used
> for synchronization and establishment of communication channels
> only. They happen to live in the file system namespace because that is
> how Unix works, but there is really no point at all to ever write them
> to disk.
> 
> Lennart
> 
Ok that is what I thought, I was just wondering if there was some
magical boot speedup.  I also run with /tmp as a tmpfs for a lot of the
same reasons.