From owner-apache@oss.sgi.com Wed Apr 5 22:22:46 2000 Received: by oss.sgi.com id ; Wed, 5 Apr 2000 22:22:27 -0700 Received: from pneumatic-tube.sgi.com ([204.94.214.22]:314 "EHLO pneumatic-tube.sgi.com") by oss.sgi.com with ESMTP id ; Wed, 5 Apr 2000 22:21:58 -0700 Received: from nodin.corp.sgi.com (nodin.corp.sgi.com [192.26.51.193]) by pneumatic-tube.sgi.com (980327.SGI.8.8.8-aspam/980310.SGI-aspam) via ESMTP id WAA09455 for ; Wed, 5 Apr 2000 22:25:45 -0700 (PDT) mail_from (mja@trudge.engr.sgi.com) Received: from trudge.engr.sgi.com (trudge.engr.sgi.com [163.154.38.51]) by nodin.corp.sgi.com (980427.SGI.8.8.8/980728.SGI.AUTOCF) via ESMTP id WAA23138 for ; Wed, 5 Apr 2000 22:21:57 -0700 (PDT) Received: (from mja@localhost) by trudge.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id WAA99154; Wed, 5 Apr 2000 22:19:24 -0700 (PDT) From: mja@trudge.engr.sgi.com (Mike Abbott) Message-Id: <200004060519.WAA99154@trudge.engr.sgi.com> Subject: Re: [Apache 10xpatch-1.3.1-2] To: ursus@usa.net (ursus) Date: Wed, 5 Apr 2000 22:19:23 -0700 (PDT) Cc: apache@oss.sgi.com In-Reply-To: <20000331185708.9015.qmail@nw173.netaddress.usa.net> from "ursus" at Mar 31, 2000 01:57:07 PM X-Mailer: ELM [version 2.5 PL2] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-apache@oss.sgi.com Precedence: bulk Return-Path: X-Orcpt: rfc822;apache-outgoing > I'm trying to use "SINGLE_LISTEN_UNSERIALIZED_ACCEPT" > ... > Is having the "UNSERIALIZED" and "FNCTL" stuff together > a bad combination? Not sure where the FNCTL is defined. SINGLE_LISTEN_UNSERIALIZED_ACCEPT and USE_FCNTL_SERIALIZED_ACCEPT can coexist. The former means it's safe for different Apache child processes to accept() simultaneously from the same socket at the same time if there is only one socket (e.g., zero or one Listen directives). The latter means, if there is more than one socket, Apache must serialize accept() calls and fcntl() is the way to do it. See htdocs/manual/misc/perf-tuning.html#acceptserialization in your patched Apache tree for an explanation of all this. > When I add "-ldl" > to the build options (EXTRA_CFLAGS) then the compilation > complains that dl isn't used for most objects, except > now httpd finally builds clean (this appears to use dl). Try moving the "-ldl" from the compiler flags (EXTRA_CFLAGS) to the list of libraries the linker needs (EXTRA_LIBS). It's really a linker option. > -D USE_QSC > -D QSC_HASH_SIZE=32768 > -D QSC_MAX_SIZE=4194304 > -D QSC_HEADER_GRAIN=32 > ... > -D USE_STAT_CACHE > -D STAT_CACHE_SIZE=32768 > -D STAT_CACHE_TIME=300 > -D STAT_CACHE_PATHLEN=128 > > My webserver has 2GByte memory (Linux 2.2.13+bigmem) > so I'm not sure whether the above QSC/STAT_CACHE > settings are appropriate? > ... > I have ~40k active objects With that much memory and that many files to cache you could stand to increase QSC_MAX_SIZE. Watch the "Memory use / total" line on the server-status page. You should probably double QSC_HASH_SIZE too, although for diminishing returns -- but again only if you're using the QSC for all of them. Read htdocs/manual/qsc.html (especially at #status) in your patched Apache tree for what all the stats mean, how to see them, and how to tune your server to make the QSC run smoother. Your STAT_CACHE_SIZE seems big. The stat cache is really meant to be small, and if most of the files are in the QSC the stat cache isn't doing much anyway. See htdocs/manual/misc/perf-mja.html#statcache The rest look OK. > Another idea I wanted to bring up is the possiblity > for an automatic or auto-adapting MMAP function aka boa > seems to have, in that you don't need to explicitly > mmapfile (which is a major pain with >40k objects, > and the object pool changing quite often.) You're not the only one. There is a tradeoff though, as always: Maintaining the cache in software slows things down. An adaptive mod_cache is available in ftp://ftp.apache.org/dist/contrib/modules/1.3/apache-contrib-1.0.8.tar.gz but I have not used it and I'm told it's buggy. > Finally is there a way to run 10x-patched Apache > with the mod_ssl add-ons? Someone on this list > mentioned they'd had "limited" success on this? A few people, including myself, have applied both the 10x patches and the mod_ssl patches. Apply 10x first, then mod_ssl. There are conflicts you must resolve manually. I don't know of a good way to help everyone who wants to use both: I don't want to distribute a combined patch because that might ruffle feathers over in mod_ssl land, but there are a couple of places in the combined tree that need special attention. Anyone have any ideas? > ursus@usa.net -- Michael J. Abbott mja@sgi.com http://reality.sgi.com/mja/ From owner-apache@oss.sgi.com Mon Apr 17 20:51:16 2000 Received: by oss.sgi.com id ; Mon, 17 Apr 2000 20:50:56 -0700 Received: from mailbox.engr.sc.edu ([129.252.22.203]:43783 "EHLO mailbox.engr.sc.edu") by oss.sgi.com with ESMTP id ; Mon, 17 Apr 2000 20:50:48 -0700 Received: by mailbox.engr.sc.edu with Internet Mail Service (5.5.2448.0) id ; Mon, 17 Apr 2000 23:48:06 -0400 Message-ID: From: "Gruen, Robert" To: "'apache@oss.sgi.com'" Subject: Accelerator Patches for Apache on FreeBSD Date: Mon, 17 Apr 2000 23:48:00 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-apache@oss.sgi.com Precedence: bulk Return-Path: X-Orcpt: rfc822;apache-outgoing Can these patches be applied to Apache for FreeBSD? From owner-apache@oss.sgi.com Tue Apr 18 17:21:22 2000 Received: by oss.sgi.com id ; Tue, 18 Apr 2000 17:21:02 -0700 Received: from deliverator.sgi.com ([204.94.214.10]:41548 "EHLO deliverator.sgi.com") by oss.sgi.com with ESMTP id ; Tue, 18 Apr 2000 17:20:46 -0700 Received: from trudge.engr.sgi.com (trudge.engr.sgi.com [163.154.38.51]) by deliverator.sgi.com (980309.SGI.8.8.8-aspam-6.2/980310.SGI-aspam) via ESMTP id RAA20409 for ; Tue, 18 Apr 2000 17:16:01 -0700 (PDT) mail_from (mja@trudge.engr.sgi.com) Received: (from mja@localhost) by trudge.engr.sgi.com (980427.SGI.8.8.8/970903.SGI.AUTOCF) id RAA56096 for apache@oss.sgi.com; Tue, 18 Apr 2000 17:19:27 -0700 (PDT) From: mja@trudge.engr.sgi.com (Mike Abbott) Message-Id: <200004190019.RAA56096@trudge.engr.sgi.com> Subject: Re: Accelerator Patches for Apache on FreeBSD To: apache@oss.sgi.com Date: Tue, 18 Apr 2000 17:19:27 -0700 (PDT) In-Reply-To: from "Gruen, Robert" at Apr 17, 2000 11:48:00 PM X-Mailer: ELM [version 2.5 PL2] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-apache@oss.sgi.com Precedence: bulk Return-Path: X-Orcpt: rfc822;apache-outgoing > Can these patches be applied to Apache for FreeBSD? http://oss.sgi.com/projects/apache/faq.html#Q7 I don't know the answer. The only major problem should be porting qsc_cas(). If you get the patches working on FreeBSD please contribute your modifications to the project so others may benefit. -- Michael J. Abbott mja@sgi.com http://reality.sgi.com/mja/