cvs commit: src/sys/conf kern.post.mk

[Wilko Bulte]

On Wed, Sep 14, 2005 at 06:39:40PM +0100, Robert Watson wrote..
> 
> On Wed, 14 Sep 2005, Marcel Moolenaar wrote:
> 
> >>>We probably ought to move this discussion to another list, but -- I 
> >>>remember two very specific occasions where I first realized how 
> >>>important an isolated /var is:
> >>>
> >>>(1) In about 1995, when I first started using ppp, I discovered the
> >>>    hard way that the default logging level was set a bit high, and 
> >>>    filled
> >>>    the entire hard disk with log records in a couple of days.
> >*snip*
> >>
> >>I've also hit cases where the log come in so fast that you can fill a
> >>1GB /var fast enough that newsyslog never has a chance to compress the
> >>log.
> >*snip*
> >
> >Just so that people realize: what is being described is not an argument 
> >for having /var be a separate partition, but really for having /var/log 
> >be a separate partition. It's just that the granularity of our thinking 
> >is highly influenced by our legacy, even to the extend of it becoming an 
> >intellectual jail.
> >
> >I think it's much easier to size a /var/log partition effectively than 
> >it is to size /, /usr and /var effectively...
> >
> >Just some food for thought,
> 
> Yeah, I specifically mentioned /var/mqueue and /var/mail as examples of 
> other components.
> 
> I agree though that what current and past state of the art has supported 
> is narrowing our thinking.  If we were using a system like AFS (listen to 
> the groans from the crowd who hates it when I harp on AFS!), we would 
> simply allocate different volumes for the directories using the same 
> back-end storage pool, and be able to administratively change the volume 
> quotas at low overhead.  The traditional BSD/UNIX quota model and/or file 
> system model has no way to express this sort of notion, and it's a very 
> useful notion.

/me goes out and dynamically expands his RAID LUN on the FC array ;)

-- 
Wilko Bulte				wilko at FreeBSD.org