Date: Wed, 26 May 1999 12:01:43 +0930 From: Greg Lehey <grog@lemis.com> To: John Reynolds~ <jreynold@sedona.ch.intel.com> Cc: freebsd-questions@FreeBSD.ORG Subject: Re: How 'tweaked' is ftp.cdrom.com ? Message-ID: <19990526120143.D667@freebie.lemis.com> In-Reply-To: <14153.54020.834915.91060@hip186.ch.intel.com>; from John Reynolds~ on Mon, May 24, 1999 at 03:30:28PM -0700 References: <14153.54020.834915.91060@hip186.ch.intel.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday, 24 May 1999 at 15:30:28 -0700, John Reynolds~ wrote: > > Hey there, don't know if this is appropriate for -questions or -chat, but it > *is* a question, so here goes: > > On slashdot.org today there was a posting about Wcarchive hitting 1.39Tb of > info in 24 hours. Yeah! :) In the comments section there were all sorts of > pseudo-FUD going around about how Linux could probably handle the same load > and blah blah blah. > > My question is--how "tweaked" is the kernel that is running ftp.cdrom.com? I > believe I've seen some postings in -current from David Greenman about > maxusers and other kernel parameters and he mentioned some settings that were > used on wcarchive. Well, how "tweaked" is that kernel vs. a "stock" 4.0-current > or 3.2-RELEASE kernel? Is it a matter of knowing what parameters to tweak in > the config file or are there lots of actual code changes? Somebody mentioned > on slashdot that that machine is actually "using" all 4Gb of RAM. Are special > boot blocks needed for this? Hmm. I had expected David to answer this one by now. AFAIK David has some special kernel code in there which doesn't work in the general case, but which improves performance in this specific application. On the other hand, he's continually feeding back improvements into the stock kernel. For example, until about 6 months ago the kernel had an address space of 256 MB (address range 0xf0000000 to 0xffffffff). This proved to be too little for very large memory systems, and it was recently changed to 1 GB (address range 0xc0000000 to 0xffffffff). On the current implementation, the rest is available for user programs, but in practice we haven't seen any processes which required 3 GB of address space. Most System V implementations give user and kernel spaces 2 GB each. Greg -- When replying to this message, please copy the original recipients. For more information, see http://www.lemis.com/questions.html See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990526120143.D667>