On Thu, Nov 19, 2009 at 7:43 PM, Aaron Mason <simplersolut...@gmail.com> wrote: > On Thu, Nov 19, 2009 at 7:57 PM, Joachim Schipper > <joac...@joachimschipper.nl> wrote: >> On Wed, Nov 18, 2009 at 12:55:03PM -0700, Bob Beck wrote: >>> 2009/11/18 Janusz Gumkowski <janusz.gumkow...@am.torun.pl>: >>> >> Is it at all possible to have more than 992 simultaneous authpf users ? >>> >> >>> > >>> >>> Yes, use more than one machine. >>> >>> > Digging out an old post of mine, still not having any real solution >>> > but a couple of ugly hacks instead, trying to get rid of them finally. >>> > >>> > To the point: is allocating a pty for authpf logins really necessary ? >>> >>> Yes. >>> >>> > What side-efects can I expect if I disable it ? >>> >>> Probably bad things. >> >> Wouldn't it be possible to crank the number of ptys? I'm by no means an >> expert, but src/sys/kern/tty_pty.c does have some interesting-looking >> #defines. (Of course, you'd also have to path libutil and who knows what >> else...) >> >> Joachim >> >> > > You'd be better off getting a second machine and CARPing them together > rather than mess with the kernel. You'd also be far more likely to > get support than if you modified the kernel (in which case you'd get > little or none I'm sure). You'd also get a degree of redundancy if > one machine bails. > > HTH > > -- > Aaron Mason - Programmer, open source addict > I've taken my software vows - for beta or for worse > >
Throwing more hardware at it can't be the real solution, not when the problem is an arbitrary system constant, and especially since the number of ptys has little to do with how many users an authpf system can support.
