Date: Fri, 7 Dec 2001 14:39:02 -0500 From: Garance A Drosihn <drosih@rpi.edu> To: Robert Watson <rwatson@FreeBSD.ORG> Cc: arch@FreeBSD.ORG Subject: Re: Default value for maxusers Message-ID: <p05101000b836c5d8cf45@[128.113.24.47]> In-Reply-To: <Pine.NEB.3.96L.1011207135049.42818N-100000@fledge.watson.org> References: <Pine.NEB.3.96L.1011207135049.42818N-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
At 1:53 PM -0500 12/7/01, Robert Watson wrote: >On Fri, 7 Dec 2001, Garance A Drosihn wrote: > >> Isn't this a case where GENERIC has a value set for "older, smaller" >> machines, so it can be used for booting up any machine? Most modern > > machines may want 64 or more, but what happens for older machines > > if we increase that value? > >Quite possibly. On the other hand, we've started to trim some older >hardware support from GENERIC over time: in the last few months, we've >dropped i386, and emulated math coprocessors. By the way, I didn't explicitly state it in my previous message, but it does seem reasonable to *me* (with my machines) that the default value should be raised, one way or another. I just don't know how raising it would effect others. Maybe raise it to some midpoint between 32 and 64? Maybe 40 or 48? The result does not have to be great on older hardware, but I do think it should be "bootable". > > Another thing I sometimes wonder is if that value (MAXUSERS) sets > > the right values for whatever it is setting. I mean, I always > > increase maxusers on my machines, but on the other hand most of my > > machines never have more than three people connected to them at > > any one time. > >Dunno. It may be that "maxusers" is simply an out-dated term, and >we should break it down into its components, seperately tweakable at >boot-time using loader.conf. Many sites already seperately define >NMBCLUSTERS to optimize network behavior independently from maxusers >and the tables it implies. Well, "maxusers" sounds like a user-friendly measurement, in the sense that it's a question that any person might feel comfortable guessing at. If we remove that setting, then people have to guess at how many NMBCLUSTERS that they need. I dread installing things which ask me questions where I don't even know what the question means, never mind what the right answer is... I'm thinking more that maybe the mapping between the MAXUSERS value and <whateverItEffects> should be different. So, we would not change the default value for MAXUSERS, but we would change how many NMBCLUSTERS are assigned per MAXUSER, and thus a default setting of 32 be reasonable for many of the machines which run freebsd. I guess the first question is "what *does* maxusers effect?", and then out of that list, which of those values are "incredibly low" for the default installation? In some sense, I would actually prefer if the actual default value for MAXUSERS was *lower* (say 16), and but that the things which key off that value were changed such that MAXUSERS=16 has the *effect* of what is now meant by (say) MAXUSERS=48. On the other hand, maybe it's less confusing to just increase the value of MAXUSERS and stick with the same mapping... :-) -- Garance Alistair Drosehn = gad@eclipse.acs.rpi.edu Senior Systems Programmer or gad@freebsd.org Rensselaer Polytechnic Institute or drosih@rpi.edu To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?p05101000b836c5d8cf45>