On Feb 25,  3:23pm, Ale Terlevich wrote:
> Subject: Re: update
>
>
> > This is a problem with the StrongArm cache-cleaning. That is where all the
CPU
> > time is going. A bug in sync was fixed a week or so ago such that it now
works
> > properly. This has resulted in a lot more cache cleaning and the
performance
> > hit you are seeing.
>
>  Ok, that's sync, but why is update using up CPU while it's meant to be
> sleeping? (Doesn't it just sleep for 30 seconds, then do a sync)

Not sure about this one. It may be something to do with the way the accounting
is done. Was the machine otherwise idle at the time? Idle time can appear in
all sorts of odd places.

> >
> > >   Are there any ramifications to running without update? (I'm using the
> > > machine mainly as an xterminal at the mo, i.e. not writing much to disk)
> >
> > Basically all update does is to sync the disks. If you don't sync the disks
> > then the cached information won't get written back. Of course unmounting
syncs
> > the disks, but if the machine crashes in the meantime you might have rather
a
> > messy fsck to do.
>
>   So does the kernel never write something to the disk simply becasue
> it's been lying in the cache for too long?

Guess what update does... :)

Cheers

Rob