Date: Thu, 6 Dec 2001 11:05:15 -0500 (EST) From: Mike Silbersack <silby@silby.com> To: Leo Bicknell <bicknell@ufp.org> Cc: "Michael R. Wayne" <wayne@staff.msen.com>, <freebsd-hackers@FreeBSD.ORG> Subject: Re: Can TCP changes be put in RELENG_4? Message-ID: <Pine.BSF.4.30.0112061058020.22493-100000@niwun.pair.com> In-Reply-To: <20011206105458.A62092@ussenterprise.ufp.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 6 Dec 2001, Leo Bicknell wrote: > Would it not be simple to create say, GENERIC-64M, GENERIC-128M, > GENERIC-256M (or small medium large, or whatever), tune a number > of critical parameters, and just ship them as part of /bin? Surely > the code to have the installer check the sysctl for the amount > of ram in the machine and hard link that one to /kernel would > be trival, and at least give us something. Just by changing a > few memory settings and making a "guess" about maxusers (and > the cascade of changes that has) based on total RAM would be a > crude but useful start. > > -- > Leo Bicknell - bicknell@ufp.org - CCIE 3440 This discussion about shipping mulitple kernels is getting silly, as nothing in this discussion has addressed why we need multiple kernels. As of 4.4, nearly every important setting can be tuned through options in loader.conf; I updated the tuning manpage to recognize this a few weeks ago. Give it a look over, I think most everything can be done with those settings. What would be interesting to know is whether or not any non-tuneable options make a performance difference. For example, I386_CPU, unneeded device drivers, etc. If that's the case, maybe multiple kernels is a possibility. However, if we're just talking about maxusers and mbufs, there's no need for a different kernel. Mike "Silby" Silbersack To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.30.0112061058020.22493-100000>