[RFC] /var versus /srv
skvidal at fedoraproject.org
Fri Sep 21 14:32:48 UTC 2007
On Fri, 2007-09-21 at 10:29 -0400, Rob Crittenden wrote:
> Matthew Miller wrote:
> > On Fri, Sep 21, 2007 at 09:19:43AM -0400, seth vidal wrote:
> >> As a sysadmin /srv is a useful thing - it's what most sysadmins do
> >> anyway - create a top level path where they mount the large, local disks
> >> and put all their data. So they know on every system if they hit /etc
> >> and /srv with the backups they'll have what they should be worried
> >> about. All admins may not call it /srv but they do something like
> >> it: /fs, /local, /data, /srv
> >> it's all the same result.
> >> so while your argument for not using it in the distro is fine -the
> >> reality is that this is what is actually done by sysadmins all over the
> >> world.
> > +1
> > Thank you Seth.
> > /var is transient data. There should be nothing there that needs backups.
> > And users shouldn't look there for files they might edit.
> Transient and not backed up? What about /var/mail, /var/spool/cron and
- /var/log - shouldn't matter - it's being sent to centralized log hosts
which I've always had put files in /srv/logs
- /var/mail has no data - all your mail should be in your central mail
server and not in /var/mail but in another path /srv/mail or /srv/mqueue
- /var/spool/cron doesn't have any files in it b/c users are not allowed
to add cron jobs except on highly specific systems. Moreover, if you're
adding root or system-controlled cron jobs they should go in /etc/cron.d
or in the /etc/cron.hourly, /etc/cron.daily, etc directories.
never in /var/spool/cron and NEVER add by such a cumbersome tool as cron
More information about the fedora-devel-list