On 07:19 Tue 22 Nov     , Heitzso wrote:
> Bruce O'Neel wrote:
> 
> >I wonder if it is worth having NKMEMPAGES_MAX set bigger in the
> >GENERIC kernel?  I suspect that most of the sparc systems
> >that are running NetBSD today have a fair bit of memory, if for
> >no other reason than it's both cheap and the upgrade is
> >worth it if you have an old system.
> >
> > 
> >
> Regarding this, some nice clean doc on how these
> variables interact would be useful.  For example, I
> ran out of mem on my ss20 w/ 512M.  Ran into both
>    NKMEMPAGES
> and
>    NKMEMPAGES_MAX
> google and various email lists. 
> 
> I bumped up NKMEMPAGES but I feel anxious
> because there's also the _MAX variant, etc.  In my
> case bumping up NKMEMPAGES but not
> NKMEMPAGES_MAX managed to up the reported
> sysctl for nkmempages and everything seems to be
> okay.  ... but what is the algorighm that controls this?
> Why would someone want to touch one and not the
> other? 
> 
> Etc.
> 
> Just trying to flag an area that is ambiguous.
> 

The real problem is that it's not obvious that this setting is related
to the other issues it causes, and so it's easy to spend a lot of time
tracking down problems that aren't there (like SCSI bus errors).

-- 
Eric Radman  |  http://eradman.com