Return-Path: Delivered-To: apmail-jakarta-tomcat-dev-archive@apache.org Received: (qmail 24434 invoked from network); 20 Jun 2002 09:56:22 -0000 Received: from unknown (HELO nagoya.betaversion.org) (192.18.49.131) by daedalus.apache.org with SMTP; 20 Jun 2002 09:56:22 -0000 Received: (qmail 27302 invoked by uid 97); 20 Jun 2002 09:56:34 -0000 Delivered-To: qmlist-jakarta-archive-tomcat-dev@jakarta.apache.org Received: (qmail 27282 invoked by uid 97); 20 Jun 2002 09:56:33 -0000 Mailing-List: contact tomcat-dev-help@jakarta.apache.org; run by ezmlm Precedence: bulk List-Unsubscribe: List-Subscribe: List-Help: List-Post: List-Id: "Tomcat Developers List" Reply-To: "Tomcat Developers List" Delivered-To: mailing list tomcat-dev@jakarta.apache.org Received: (qmail 27270 invoked by uid 98); 20 Jun 2002 09:56:33 -0000 X-Antivirus: nagoya (v4198 created Apr 24 2002) content-class: urn:content-classes:message MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Subject: RE: mod_jk 1.2.0 release news X-MimeOLE: Produced By Microsoft Exchange V6.0.5762.3 Date: Thu, 20 Jun 2002 11:56:47 +0200 Message-ID: X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: mod_jk 1.2.0 release news Thread-Index: AcIYO6T+3TwxnIAsQeGw0W8Cx842AAABM3YA From: "GOMEZ Henri" To: "Tomcat Developers List" X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N Good catch Mladen. I think we may be near the solution, since after tunning server.xml for tomcat 3.3.1, i see no more error in mod_jk.log (recv failed for example), but still got error (failed request), so the error is in httpd 2.0 (and mod_jk), no more in ajp13 (at least for tc 3.3.1) - Henri Gomez ___[_]____ EMAIL : hgomez@slib.fr (. .) =20 PGP KEY : 697ECEDD ...oOOo..(_)..oOOo... PGP Fingerprint : 9DF8 1EA8 ED53 2F39 DC9B 904A 364F 80E6=20 >-----Original Message----- >From: Mladen Turk [mailto:mturk@mappingsoft.com] >Sent: Thursday, June 20, 2002 11:19 AM >To: 'Tomcat Developers List' >Subject: RE: mod_jk 1.2.0 release news > > >> From: GOMEZ Henri >>=20 >> Ok, since it appears we still have problems under heavy >> load when using Apache 2.0 (with TC 3.3.1/4.x) or=20 >> with Apache 1.3 + Tomcat 4.x, I'll delay the 1.2.0 >> until we fix that problem. >>=20 >> I'll works on 2 parts : >>=20 >> 1) Fix the TC 4.x problems with ajp13 (jtc) >> 2) Fix the Apache 2.0 problems >>=20 >> For so I'll need some help, especially in TC 4.x area >> to determine what's going wrong with TC4 ajp implementation. >>=20 > >OK here are some of my experiments and results: > >1. The major problem IMO are the multiple threads that are trying to >acquire the sockets. > >To be able to find wtf is going on I've basically made the >jk_open_socket to be the single threaded. I've made the single static >socket, and checking if the connect() retuns EISCONN, and if=20 >do, sleep a >while and then try again, until the 20 sec timeout. > >And here are the results: > >Concurrency Level: 2 >Time taken for tests: 20.151314 seconds >Complete requests: 2000 >Failed requests: 0 >Write errors: 0 >Keep-Alive requests: 0 > >Concurrency Level: 4 >Time taken for tests: 20.161001 seconds >Complete requests: 2000 >Failed requests: 0 >Write errors: 0 >Keep-Alive requests: 0 > >The results looks excellent at first, but as you can guess, the tests >are finishing around 20 secs and the timeout is set to the 20 sec. When >having two worker threads active I'm facing the 1 timeout and=20 >that on is >for the second!(remember that) connection (Actually there=20 >should be two, >but the second is probably lost cause I'm getting the ENOTSOCK for >connect in the mod_jk.log, on test end). > >No matter how many connections there are the second (per-thread) >connection that is made to TC gets timed-out. > >2. The reason? IMO we are not closing the first connection, or there is >a bug when reading from the connection for the first time. > >3. Cause of that when not having single threaded jk_open_socket the >first connection on each thread is never closed and as apache recycles >its threads, we finally end up with the 2000 or more opened sockets per >10000 requests. > > > >MT. > > >-- >To unsubscribe, e-mail: =20 For additional commands, e-mail: = -- To unsubscribe, e-mail: For additional commands, e-mail: