| 03:23.11 | *** join/#tomcat kousuke (n=kikuchi@221x251x19x21.ap221.ftth.ucom.ne.jp) |
| 03:33.49 | *** join/#tomcat jjchen (n=jjchen@61.171.10.195) |
| 05:21.48 | *** join/#tomcat jasonb_ (i=noneoyer@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 07:07.02 | *** join/#tomcat CrummyGummy (n=CrummyGu@dsl-242-57-71.telkomadsl.co.za) |
| 07:39.15 | *** join/#tomcat fix_ (n=fix@cable-87-244-191-89.upc.chello.be) |
| 07:42.24 | *** join/#tomcat [algo] (n=ilia@algo.dialup.corbina.ru) |
| 08:11.13 | *** join/#tomcat caotic (n=caotic@189.157.116.97) |
| 08:18.14 | *** join/#tomcat gregor_k (n=a@p54a1b279.dip0.t-ipconnect.de) |
| 08:32.38 | *** join/#tomcat fix_ (n=fix@cable-87-244-191-89.upc.chello.be) |
| 08:51.41 | *** join/#tomcat stork (n=sh2-136@storkey.plus.com) |
| 09:04.18 | *** join/#tomcat seshf (i=kvirc@mx1.hesgdv.de) |
| 09:06.42 | *** join/#tomcat columbo (n=Miranda@pool-71-104-52-62.lsanca.dsl-w.verizon.net) |
| 10:46.28 | *** join/#tomcat guscel (n=chatzill@www.heads.com) |
| 10:58.59 | gustafc | Hi, I'm running TC 6.0.10 (on win vista/java 6) and I'd like to map a context to a directory outside of catalina_home (tp a dirextory in my Eclipse workspace so I won't have to rebuild the war every time I change something). The directory has a path like C:\path\to\WebContent and is a kosher webapp directory (complete with WEB-INF directory etc). I want to map it so that when I go to e.g.... |
| 10:59.01 | gustafc | ...localhost:8080/test/index.html I'll see C:\path\to\WebContent\index.html. I tried adding the following to my server.xml: <Context path="/test" docBase="C:\path\to\WebContent\" reloadable="true" /> but it doesn't seem to work (I probably got it wrong, silly me). Any leads? I tried symlinking it into CATALINA_HOME\test with mklink but it seems TC didn't like/understand the symlink. |
| 11:07.44 | gustafc | Correction: I tried symlinking it into CATALINA_HOME\webapps\test |
| 11:34.16 | [algo] | how can I make Tomcat serve static context ? |
| 11:36.06 | [algo] | version is 6.0.10 |
| 11:36.11 | [algo] | OS is Linux Gentoo |
| 11:36.18 | [algo] | kernel 2.6.19 |
| 11:44.46 | *** join/#tomcat space (n=space@62.193.9.2) |
| 11:45.25 | space | Is mod_jk2 deprecated ? |
| 11:46.43 | space | This is the link told about it : http://wiki.apache.org/tomcat/Tomcat/Jk2Connector |
| 12:05.15 | *** part/#tomcat space (n=space@62.193.9.2) |
| 12:15.31 | *** join/#tomcat randrew (n=raj@dolmen.cc.columbia.edu) |
| 12:20.35 | *** part/#tomcat randrew (n=raj@dolmen.cc.columbia.edu) |
| 12:22.14 | *** join/#tomcat randrew (n=raj@dolmen.cc.columbia.edu) |
| 14:22.26 | *** join/#tomcat eddon (n=e@p549F11D4.dip0.t-ipconnect.de) |
| 14:33.25 | *** join/#tomcat a4akb (n=a4akb@62.215.156.215) |
| 14:33.52 | a4akb | yo |
| 14:44.40 | *** join/#tomcat maier (n=maier@p14-127.med.unc.edu) |
| 15:17.10 | *** join/#tomcat CrummyGummy (n=CrummyGu@dsl-242-57-71.telkomadsl.co.za) |
| 15:24.48 | *** join/#tomcat jasonb (i=noneoyer@adsl-66-124-73-250.dsl.sntc01.pacbell.net) |
| 15:38.08 | *** join/#tomcat maier (n=maier@p14-127.med.unc.edu) |
| 15:48.04 | *** join/#tomcat kab (n=abarrera@148.229.1.11) |
| 15:56.52 | *** join/#tomcat roue (n=roue@afterthought.ACS.Carleton.edu) |
| 16:26.26 | *** join/#tomcat compair1 (i=ariel@gateway/tor/x-05fb16765a5d0ad6) |
| 17:04.23 | *** join/#tomcat Hellaenergy (n=Hellaene@66.187.199.82) |
| 17:15.10 | a4akb | Hellaenergy: hello |
| 17:26.40 | *** join/#tomcat pappy-_ (n=alex@dslb-084-058-209-044.pools.arcor-ip.net) |
| 17:30.53 | *** join/#tomcat el_gato_malo (n=gschuett@4.38.41.141) |
| 17:30.56 | el_gato_malo | ah good sires |
| 17:31.36 | el_gato_malo | can some one tell me how to fix a SEVERE: All threads (500) are currently busy, waiting. Increase maxThreads (500) or check the servlet status |
| 17:31.36 | el_gato_malo | ~ error in tomcat 5.5.17 on redhat with 2.6.9 kernel jvm 1.5 sun |
| 17:32.03 | el_gato_malo | should there not be a thread timeout or something I see 20000 as a setting in server.xml |
| 17:33.53 | el_gato_malo | I run a movie download service |
| 17:34.00 | el_gato_malo | I wish I had 500 customers downloading at once |
| 17:34.07 | el_gato_malo | so I suspect some bs configuration |
| 17:34.17 | el_gato_malo | I have 18 other servers with identical config workign fine |
| 17:34.28 | el_gato_malo | I wonder whats gumming up 500 threads? |
| 17:34.45 | el_gato_malo | I suspect some java incompetence or perhaps apache/tomcat problem |
| 17:34.47 | el_gato_malo | hm |
| 17:38.07 | el_gato_malo | why would threads be busy |
| 17:38.08 | el_gato_malo | hm |
| 17:38.11 | el_gato_malo | f--k |
| 17:39.00 | *** join/#tomcat lintlock (n=look@cpe-071-076-223-082.triad.res.rr.com) |
| 17:39.05 | lintlock | ping jasonb |
| 17:39.17 | lintlock | :P |
| 17:43.50 | el_gato_malo | hey JASON |
| 17:44.18 | el_gato_malo | lol |
| 17:44.21 | pfn | el_gato_malo, threads would be busy because your application is b0rked |
| 17:44.29 | el_gato_malo | oh? |
| 17:44.40 | el_gato_malo | but whys it happy on liek 14 other identical linux boxen |
| 17:44.42 | el_gato_malo | hm |
| 17:44.52 | el_gato_malo | our pogrammers are kinda toolish |
| 17:45.00 | el_gato_malo | I even did a diff on server.xml |
| 17:45.02 | el_gato_malo | no diff |
| 17:45.03 | el_gato_malo | :( |
| 17:45.11 | pfn | el_gato_malo, then it's the software |
| 17:45.20 | pfn | el_gato_malo, the answer is simple, invoke a thread stack dump on tomcat |
| 17:45.27 | pfn | el_gato_malo, or use jconsole if you have jmx enabled |
| 17:45.38 | pfn | (the latter being much easier to use) |
| 17:45.46 | pfn | you can look at each thread and see what it's doing |
| 17:45.54 | pfn | and even give that to the toolish developers to fix |
| 17:47.37 | lintlock | pfn you were saying at one point i could use request.setAttribute("somename",object) |
| 17:47.50 | pfn | yes |
| 17:47.56 | lintlock | if that object is an array list would i just use the methods of an array list in the foreach? |
| 17:48.05 | lintlock | like arraylist.get(num); |
| 17:48.24 | pfn | no |
| 17:48.34 | pfn | assume somename = someArrayList |
| 17:48.43 | lintlock | ok |
| 17:48.48 | pfn | forEach var="arrayItem" items="${someArrayList}" |
| 17:49.06 | pfn | each arrayItem in the loop is each element in the list |
| 17:49.30 | lintlock | so i would use a method of that class i have implemented |
| 17:49.36 | lintlock | like getID |
| 17:49.45 | pfn | you would use arrayItem.ID |
| 17:50.32 | lintlock | hummmm |
| 17:50.33 | el_gato_malo | thank you |
| 17:50.59 | lintlock | does ID have to be a public variable? |
| 17:52.12 | pfn | no, not a public variable |
| 17:52.26 | pfn | any .property access on a variable in EL *must* be a javabean property |
| 17:52.32 | pfn | thus, you must have a getter |
| 17:53.10 | *** join/#tomcat vinse (n=vinse@208.253.223.146) |
| 17:53.19 | lintlock | ok |
| 17:53.29 | lintlock | so arrayItem.getID |
| 17:53.30 | *** part/#tomcat randrew (n=raj@dolmen.cc.columbia.edu) |
| 17:53.42 | lintlock | thats the name of my getter |
| 17:53.49 | pfn | no, then arrayItem.ID |
| 17:53.52 | pfn | is the name of the property |
| 17:54.14 | pfn | EL: arrayItem.ID = Java: arrayItem.getId() |
| 17:54.54 | lintlock | got ya |
| 17:55.12 | pfn | er, .getID() |
| 17:55.15 | pfn | but you get my point |
| 17:56.00 | lintlock | yep |
| 17:56.09 | lintlock | time to rewrite this page |
| 17:57.24 | lintlock | i had a scriptlet not work on another computer but work on mine |
| 17:57.31 | lintlock | so i had no choice but to change the format |
| 18:10.14 | el_gato_malo | heh |
| 18:10.23 | el_gato_malo | how do I cause a thread stack dump again? |
| 18:10.24 | el_gato_malo | :) |
| 18:10.27 | el_gato_malo | java 1.5 |
| 18:10.31 | el_gato_malo | tomcat 5.5.17 |
| 18:10.33 | el_gato_malo | on linux |
| 18:10.37 | el_gato_malo | redhat 2.6.9 |
| 18:10.42 | el_gato_malo | kernel |
| 18:16.22 | Marv|LG | kill -QUIT <pid> |
| 18:25.42 | *** join/#tomcat vinse (n=vinse@208.253.223.146) |
| 18:29.07 | *** join/#tomcat JSPJunkie (n=DSCowboy@ool-45717b01.dyn.optonline.net) |
| 18:44.44 | el_gato_malo | oh? |
| 18:57.53 | *** join/#tomcat Mandarino (n=Mandarin@202.pool85-50-102.dynamic.orange.es) |
| 18:57.58 | Mandarino | Hi ppl |
| 19:00.40 | JSPJunkie | http://en.wikipedia.org/wiki/Cool_Devices#Operation_07:_Yellow_Star |
| 19:10.34 | *** join/#tomcat puff` (n=user@70.90.236.161) |
| 19:15.10 | *** part/#tomcat Mandarino (n=Mandarin@202.pool85-50-102.dynamic.orange.es) |
| 19:19.52 | *** join/#tomcat Azzkikr_ (n=krpors@a62-251-82-134.adsl.xs4all.nl) |
| 19:25.38 | lintlock | i enjoy mismapping servlets |
| 19:25.44 | lintlock | damn that was stupid |
| 19:34.41 | jasonb | lintlock: Hi there. |
| 19:36.23 | lintlock | :) |
| 19:36.35 | lintlock | been working to convert my scriplets into real servlets |
| 19:39.10 | jasonb | lintlock: You might also consider making some of them into tag library tags. |
| 19:40.17 | lintlock | ummm |
| 19:40.40 | lintlock | heres a question for you |
| 19:40.40 | lintlock | i have this class http://papernapkin.org/pastebin/view/5903 |
| 19:40.45 | lintlock | an array of that class to be specific |
| 19:40.57 | lintlock | there is a foreach to go through the loop |
| 19:42.01 | lintlock | there are arraylists for some of the values of that class |
| 19:42.11 | lintlock | do i foreach each of those classes? |
| 19:42.53 | vinse | ? |
| 19:43.07 | vinse | oh you mean because you want to iterate through both collections in one forEach? |
| 19:44.07 | vinse | if that's what you mean, forEach through one, and use varStatus.index to access the corresponding members of the other collection, e.g. ${collection2[varStatus.index]} |
| 19:45.07 | lintlock | i just did a foreach and tried to access the variable |
| 19:45.26 | lintlock | that is an arraylist |
| 19:45.46 | lintlock | its getter is called getQuestion and getAnswer |
| 19:45.49 | lintlock | both return arraylist |
| 19:45.57 | vinse | lintlock: right, two different lists |
| 19:46.04 | vinse | lintlock: and you want to do what with them? |
| 19:46.14 | lintlock | just do a forloop and display them |
| 19:46.22 | vinse | in one loop? |
| 19:46.34 | lintlock | no multiple |
| 19:46.40 | lintlock | first loop to go through the class |
| 19:46.56 | vinse | what class? |
| 19:47.05 | lintlock | http://papernapkin.org/pastebin/view/5903 |
| 19:47.06 | lintlock | this one |
| 19:47.13 | lintlock | its an array of that class |
| 19:47.14 | vinse | so you have a collection of those too? |
| 19:47.18 | lintlock | QueryQuestionResults |
| 19:47.22 | lintlock | yes |
| 19:47.29 | lintlock | i get it from the previous servlet page |
| 19:47.40 | vinse | so you need a nested forEach |
| 19:47.45 | lintlock | yes |
| 19:47.54 | lintlock | i did that part |
| 19:48.08 | lintlock | i just want to make sure im accessing the variables properly |
| 19:48.09 | vinse | your question grows increasingly unclear ... |
| 19:48.31 | lintlock | i guess showing you my code would make more sense |
| 19:48.39 | vinse | did you try it? |
| 19:48.43 | vinse | like, does it not work? |
| 19:49.03 | vinse | asking about the specific thing that doesnt work would make more sense |
| 19:49.11 | lintlock | http://papernapkin.org/pastebin/view/5904 |
| 19:49.19 | vinse | and that code doesnt work? |
| 19:49.36 | lintlock | no its just im making sure im accessing those variables within the class right |
| 19:50.36 | vinse | ogm |
| 19:50.57 | *** join/#tomcat LongBeach (n=mike@AFontenayssB-152-1-78-35.w86-218.abo.wanadoo.fr) |
| 19:51.02 | vinse | so you wrote it and asked here *before* seeing if it worked? |
| 19:51.10 | vinse | no offense, but you're using irc wrong ;p |
| 19:51.23 | lintlock | no i was trying to understand what im suppose to be doing |
| 19:51.31 | lintlock | i am kind of lost |
| 19:51.40 | lintlock | i see what pfn has done |
| 19:52.06 | lintlock | but i have a class that contains variables and 2 different list that all have getters and setters |
| 19:52.36 | vinse | ok |
| 19:53.01 | vinse | first problem is you cant use varStatus directly, you have to assign in in the start of the forEach |
| 19:53.35 | vinse | <c:forEach items="${foo}" var="bar" varStatus="someName"> |
| 19:53.40 | lintlock | got ya |
| 19:54.27 | vinse | also your code there looks like it's only gonna do anything on the first iteration of the outer loop |
| 19:54.31 | vinse | is that your intention? |
| 19:54.37 | lintlock | that was the point on that one |
| 19:55.22 | lintlock | Thank you so much |
| 19:55.41 | vinse | np |
| 19:55.53 | lintlock | that code worked |
| 19:56.37 | lintlock | i am converting scriplets |
| 19:56.45 | lintlock | and its been taking awhile |
| 19:58.47 | lintlock | vinse you have helped greatly |
| 20:10.59 | *** join/#tomcat ChocoTaco (n=chatzill@wsip-64-58-175-215.oc.oc.cox.net) |
| 20:14.12 | el_gato_malo | well |
| 20:14.12 | el_gato_malo | I got buttslammed |
| 20:14.15 | el_gato_malo | for not knowing perl |
| 20:14.20 | el_gato_malo | in a sysadmin interview |
| 20:14.59 | ChocoTaco | Tomcat 5.5, Sun JDK 1.5.0.10, Debian Sarge - I installed tomcat, and when I place the sample.war in my webapps directory, the sample page comes up fine. Trying to install axis just gets me a 404, though. |
| 20:16.09 | ChocoTaco | Am I missing some step needed for deploying axis? I also tried to place a Context element in the server.xml and it didn't help anything. |
| 20:16.29 | el_gato_malo | axis? |
| 20:16.33 | el_gato_malo | apache axis? |
| 20:16.37 | ChocoTaco | yes, apache axis |
| 20:16.45 | el_gato_malo | isnt that a perl app? |
| 20:16.51 | ChocoTaco | just trying to get the axis welcome doc to come up |
| 20:17.06 | el_gato_malo | did ya read the install README? |
| 20:17.25 | ChocoTaco | nope, I have it as a java web app; I believe it's also available as a C++ cgi though |
| 20:17.39 | ChocoTaco | I read the install guide... they just sort of expect it to be working at this point |
| 20:22.49 | ChocoTaco | well there apparently is a much newer version of axis out, so I'll try that. The apache project pages for axis could sure use some updating. |
| 20:27.30 | *** join/#tomcat puff`` (n=user@c-24-131-235-208.hsd1.pa.comcast.net) |
| 20:43.36 | el_gato_malo | heh |
| 20:43.38 | el_gato_malo | amen brother |
| 20:43.44 | el_gato_malo | try the tomcat docs sometime lol |
| 20:45.58 | el_gato_malo | is tomcat fine for a webserver because any lame webserver can saturate a t1? |
| 20:46.04 | el_gato_malo | lie if I have huge photo files |
| 20:46.17 | el_gato_malo | can I server then offa tomcat as opposed to apache and it be same speed? |
| 20:46.20 | el_gato_malo | or will speed suck |
| 20:46.27 | el_gato_malo | isn't network the bottleneck? |
| 20:46.29 | el_gato_malo | or no? |
| 20:46.30 | el_gato_malo | :) |
| 20:48.44 | *** part/#tomcat JSPJunkie (n=DSCowboy@ool-45717b01.dyn.optonline.net) |
| 21:21.55 | lintlock | ping: jasonb |
| 21:22.37 | jasonb | lintlock: pong. :) |
| 21:34.03 | *** join/#tomcat student_ (n=student@stf164221.tru.ca) |
| 21:35.02 | student_ | Hello! I'm running Tomcat 5.5.23 on ArchLinux. I was wondering how to change the user launching tomcat on startup from 'tomcat' to 'somethingelse' ? Where is this preference set? |
| 21:46.34 | pfn | in your init script |
| 21:46.51 | pfn | jasonb do you know the syntax to "ProxyPass /~* !" for mod_proxy? |
| 21:47.00 | pfn | jasonb, I want tomcat to not handle ~user directories.... |
| 21:47.39 | jasonb | Yeah. |
| 21:47.47 | jasonb | ! means don't proxy it, I believe. |
| 21:47.54 | pfn | no, I mean, how do I do that |
| 21:48.01 | pfn | since apache doesn't honor ProxyPass /~ ! |
| 21:48.12 | pfn | so... I have to hardcode it per user :( |
| 21:48.19 | jasonb | Well, first, did you put a space between the ~ and the !? |
| 21:48.37 | lintlock | take care jason and pfn you guys have helped alot |
| 21:48.40 | pfn | jasonb of course |
| 21:48.43 | pfn | lintlock, where's my ++ :p |
| 21:48.54 | jasonb | See ya linty. :) |
| 21:49.15 | student_ | pfn: if I change '-user tomcat \' to '-user dspace \' in /etc/rc.d/tomcat, localhost:8080 sends me to a blank page instead of tomcat admin.. Is this not the variable to set? |
| 21:50.01 | jasonb | pfn: did you try ProxyPass /\~ ! ? |
| 21:55.17 | pfn | jasonb, ah, no, haven't tried that |
| 21:55.19 | pfn | I'll try in a bit |
| 21:56.32 | jasonb | I think it's a regular expression. |
| 21:56.39 | jasonb | So, ~ is a special character. |
| 21:56.46 | jasonb | (which should be escaped, if so) |
| 22:00.13 | pfn | nope, doesn't work |
| 22:00.18 | pfn | ProxyPass /~pfnguyen ! works |
| 22:02.04 | el_gato_malo | server.xml perhaps |
| 22:02.08 | el_gato_malo | :) |
| 22:02.16 | el_gato_malo | ps, tomcat 6 is out |
| 22:02.20 | el_gato_malo | just download that |
| 22:02.27 | el_gato_malo | why arch is back on 5.5 Ill never know |
| 22:02.29 | el_gato_malo | :( |
| 22:02.33 | jasonb | pfn: How about ProxyPass /~.* ! ? |
| 22:08.30 | pfn | jasonb nothing in the mod_proxy docs seems to indicate that ProxyPass takes a regex |
| 22:09.07 | jasonb | pfn: Well, it takes a path, and lots of paths in the httpd conf are actually regexes. |
| 22:09.15 | pfn | jasonb, and nope, that doesn't work, either |
| 22:09.29 | pfn | fortunately, mine is the only userdir I care about |
| 22:09.29 | pfn | heh |
| 22:09.38 | pfn | so this isn't such a big deal |
| 22:10.19 | jasonb | pfn: I think your problem might be that you can specify a path prefix as long as it either ends with a slash, or a slash would normally immediately follow the prefix you specify. But, you want to end with the ~ because the thing after that is what varies. |
| 22:10.31 | pfn | jasonb indeed |
| 22:10.38 | jasonb | pfn: You can, however accomplish what you want by using mod_rewrite. |
| 22:11.05 | pfn | I don't see how... |
| 22:11.20 | jasonb | That really does take a regex. |
| 22:11.33 | pfn | yeah, but I don't see how that can be used in conjunction with ProxyPass to make it work |
| 22:11.44 | pfn | unless mod_rewrite can make proxypass rules "not hit" |
| 22:13.59 | jasonb | RewriteRule ^/~.*$ http://your.server.dom/error-page.html [R,L] |
| 22:14.12 | jasonb | Put that above the config for mod_proxy. |
| 22:14.54 | pfn | well, of course that will work, since it's a redirect and occurs before proxy config |
| 22:15.00 | jasonb | Yup. |
| 22:15.19 | pfn | that's still a few steps short of "noproxypass" for userdirs |
| 22:15.57 | jasonb | Don't forget the RewriteEngine On above the RewriteRule. |
| 22:16.21 | jasonb | What exactly are you trying to do with the ~user requests? |
| 22:17.17 | pfn | not proxy them to tomcat |
| 22:17.25 | pfn | i.e. leave it to apache to handle |
| 22:19.32 | jasonb | There are mod_rewrite directives that can skip other handlers. |
| 22:20.43 | jasonb | Did you also try ProxyPass /~* ! ? |
| 22:21.16 | pfn | yes |
| 22:21.20 | jasonb | Probably won't work, but it's worth trying. :) |
| 22:21.22 | *** join/#tomcat rahul_ (n=Rahul@125-237-72-70.jetstream.xtra.co.nz) |
| 22:21.27 | jasonb | Okay, then mod_rewrite. |
| 22:22.51 | *** join/#tomcat maier (n=maier@cpe-065-190-130-225.nc.res.rr.com) |
| 22:22.52 | pfn | mod_rewrites has no provisions for this... |
| 22:23.38 | *** part/#tomcat rahul_ (n=Rahul@125-237-72-70.jetstream.xtra.co.nz) |
| 22:24.50 | jasonb | H=Content-handler' (force Content handler) |
| 22:25.06 | jasonb | That might be a way.. though I've never used it. |
| 22:25.45 | pfn | I don't think that'd work or be desirable... unless there's a handler that's 'use-apache' |
| 22:25.48 | pfn | effectively |
| 22:26.24 | jasonb | Well, there is likely to be a file/directory handler. |
| 22:26.35 | jasonb | I don't know where there is a list of handler names. |
| 22:36.11 | jasonb | Another way appears to be to use PT, since then you can rewrite the /~ to /home/, and then you can tell ProxyPass not to proxy that. |
| 22:36.43 | pfn | yeah, then I have to reconfigure userdir to use /home/ or something |
| 22:37.18 | jasonb | Well, wherever your homes are, put that in the rewrite rule. |
| 22:38.19 | pfn | I guess that's probably going to be the most reasonable solution |
| 22:40.41 | pfn | excellent, got a websphere l3 contact for jsp |
| 22:40.46 | pfn | oops, mischan |
| 22:42.44 | jasonb | mmmm, websFear. :) |
| 22:49.13 | pfn | yeah, I hate it |
| 22:49.17 | pfn | it doesn't get anything right... |
| 22:49.26 | pfn | :( |
| 22:49.42 | pfn | there's at least one feature I could name in tomcat that websphere could definitely use |
| 22:49.59 | pfn | (defining different realms for different 'contexts') |
| 22:50.06 | pfn | and lacks... |
| 23:03.00 | *** join/#tomcat columbo (n=Miranda@pool-71-104-52-62.lsanca.dsl-w.verizon.net) |
| 23:28.50 | *** join/#tomcat ramza3 (n=ramza3@c-24-98-96-151.hsd1.ga.comcast.net) |
| 23:34.08 | *** join/#tomcat single_LA_CA (n=g@4.38.41.141) |
| 23:34.38 | single_LA_CA | can someone give me some fodder to convince managers that I don't need apache? I can run tomcat alone? even on front end as a proxy? |
| 23:35.00 | single_LA_CA | it is my understadning that the network is always the limiting factor! |
| 23:35.07 | single_LA_CA | am I wrong? |
| 23:38.04 | jasonb | You're not wrong. |
| 23:38.34 | jasonb | The network is almost always the limiting factor.. or other servers that your webapp may depend upon. |
| 23:38.54 | jasonb | Tomcat does not itself contain code to proxy requests to another server, but you may include 3rd party code in your webapp to do that. |
| 23:39.25 | jasonb | That's really one of Tomcat's weak points as a first-contact web server -- it cannot proxy to other servers without a webapp containing code to do it. |
| 23:39.38 | jasonb | But, such 3rd party code does indeed exist. |
| 23:40.31 | jasonb | If you mainly serve requests that must run in Tomcat, then the Apache httpd --> Tomcat setup is bad because httpd slows Tomcat down. |
| 23:40.56 | jasonb | The Tomcat requests can run much faster if httpd is not in front of Tomcat. |
| 23:44.00 | pfn | it isn't that httpd slows tomcat down, as tomcat's performance doesn't change, it's just httpd's proxying/request dispatching is slow |
| 23:44.21 | pfn | jasonb, have you tried benchmarks using something else, like faban? instead of 'ab', to get a fairer comparison? |
| 23:45.21 | jasonb | pfn: I'm using less technical language to get my point across, yes. :) |
| 23:45.31 | jasonb | pfn: Tell me what is unfair about using ab. |
| 23:46.05 | pfn | jasonb it's 'single-threaded' so it doesn't properly represent concurrent work |
| 23:46.06 | jasonb | (and don't just point me to Scott Oak's page) |
| 23:47.25 | pfn | and yeah, I guess that points to scott oak's page |
| 23:47.46 | jasonb | pfn: Well, but I'm sure you are familiar with the concept that there is little actual difference between a nicely written single thread selector and a nicely written multithreaded application, running on a single-CPU machine. |
| 23:48.09 | jasonb | I disagree with Scott Oaks on this one, with certain qualifications. |
| 23:48.22 | jasonb | If the client is not network nor CPU bound, then he's wrong. |
| 23:48.35 | pfn | I suppose you're right on that |
| 23:48.41 | jasonb | In my benchmarks, the client was neither network nor CPU bound. |
| 23:48.55 | jasonb | I kept my benchmarks very fair. |
| 23:49.44 | jasonb | Everything is single-threaded when you run software on a machine with just one processor. :) |
| 23:49.57 | pfn | yes, that's true |
| 23:50.40 | *** join/#tomcat maier (n=maier@cpe-065-190-130-225.nc.res.rr.com) |
| 23:50.59 | pfn | but then that leads up to another question |
| 23:51.04 | jasonb | Anyway, with one client thread that isn't bottlenecked, I could see different web server configurations chewing through more or less requests per second, and I never found an upper limit where the client was either CPU bound or network bound. |
| 23:51.16 | pfn | if the tester and the testee are both on the same machine, how are you certain something isn't processor bound |
| 23:51.37 | pfn | of course, since you're comparing ab vs. tomcat and vs. httpd in the same configuration |
| 23:51.48 | pfn | it should be the relative numbers that are important, rather than absolute |
| 23:52.16 | jasonb | The test client should not be running on the same machine as the server software being tested. |
| 23:52.33 | jasonb | In my benchmarks, I always had two separate machines, separated by gigabit ethernet. |
| 23:52.59 | pfn | I see |
| 23:53.03 | jasonb | I did approach network saturation on some tests where the response was largest. |
| 23:53.21 | jasonb | But, I made sure I never reached network saturation, by lowering the response size. |