Re: Findings / Last Call for Help

[rogue_(_at_)_ngxweb_(_dot_)_com]

Hi Chris,

Thanks for the reply.

I did read that section of the FAQ and I had considered it.  I still may 
give it a try.

The reason I havent tried it was because I considered it more of a 
desperate measure.  That is, I hopeful that I would find an safer 
alternative.  Given the fact that I am in So. California and we are in the 
midst of an energy shortage which will definitely produce rolling blackouts 
this summer and the fact that the isp is not using UPSs on their linux game 
servers...(they feel if it crashes, they will just reinstall if need be, 
whereas the bread and butter servers are on UPSs), I wanted to find another 
alternative in light of the danger using async poses...but, I guess 
isolating async to one small partition wouldnt be so bad.  Ill give it a try.

If you have any further suggestions, I would be happy to hear them.

Thanks!
-RS-

man I need to get some sleep hehe

At 05:32 PM 4/24/01, Chris Jones wrote:


> rogue_(_at_)_ngxweb_(_dot_)_com wrote:
> > The point of this message is to find out if there is anything I can do to
> > make OBSD faster in changing maps.
> >
> > I have enabled softupdates, enabled dummy_nops in the kernel...stripped out
> > any unneeded drivers/options (NFS, quotas, etc.) and from what I
> > understand, dirprefs is enabled by default.
> >
> > I have searched for and read all documentation on performance tunning that
> > I could find...with that in mind, the I would be happy to hear any
> > suggestions anyone may have.
>
>         Have you tried mounting the drive asychronously?  Linux (and possibly
> FreeBSD) mounts drives asynchronously, while OpenBSD mounts
> synchronously.  This slows down disk IO in exchange for increasing
> recoverability.  See FAQ 11.5
> (http://www.openbsd.org/faq/faq11.html#11.5).
>
>         I'd suggest setting aside a partition for the maps, or (if you can't
> get just the maps on there) the whole HL/CS data, and mounting that
> partition (and only that partition) asychronously.