17 messages in org.apache.tomcat.usersRe: Best Practices Question| From | Sent On | Attachments |
|---|---|---|
| ach...@saysit.com | Sep 29, 2002 1:44 am | |
| ach...@saysit.com | Sep 29, 2002 4:20 am | |
| ach...@saysit.com | Sep 29, 2002 4:23 am | |
| ach...@saysit.com | Sep 29, 2002 4:26 am | |
| Barry Moore | Sep 29, 2002 3:39 pm | |
| V. Cekvenich | Sep 29, 2002 5:31 pm | |
| Kent Perrier | Sep 29, 2002 6:07 pm | |
| Craig R. McClanahan | Sep 29, 2002 6:39 pm | |
| Oskar Bartenstein | Sep 29, 2002 8:30 pm | |
| Craig R. McClanahan | Sep 29, 2002 10:17 pm | |
| Oskar Bartenstein | Sep 30, 2002 1:10 am | |
| Turner, John | Sep 30, 2002 5:35 am | |
| Turner, John | Sep 30, 2002 5:38 am | |
| Step...@bmwfin.com | Sep 30, 2002 5:38 am | |
| Craig R. McClanahan | Sep 30, 2002 9:22 am | |
| Glenn Nielsen | Sep 30, 2002 1:13 pm | |
| Glenn Nielsen | Sep 30, 2002 1:45 pm |
| Subject: | Re: Best Practices Question | |
|---|---|---|
| From: | Oskar Bartenstein (osk...@ifcomputer.co.jp) | |
| Date: | Sep 29, 2002 8:30:18 pm | |
| List: | org.apache.tomcat.users | |
Sun, 29 Sep 2002 18:39:54 -0700 (PDT) "Craig R. McClanahan"
<crai...@apache.org> said:
IMHO, anyone who goes to all the extra effort of configuring Apache+Tomcat, instead of Tomcat alone, is nuts unless they need it. Valid reasons to need it include:
I am pondering one more: reducing interruption of service.
Tomcat can handle many webapps with ease, and you can allow as many threads as you need to serve your users.
But when a single servlet gets stuck, you have to restart the whole Tomcat machinery, taking down all running webapps, possibly loosing session data and certainly interrupting service to all users, including those completely unrelated to the hanging servlet. My experience is based on Tomcat 4.0.3 and 3.x only. I searched quite a while but failed to find a way to monitor and catch individual runaway servlet threads. Please correct me if I am wrong.
In that respect, with due appreciation for this great work, I think standalone Tomcat does not scale well.
To avoid this, Apache just working as a proxy on port 80 for many independent instances of Tomcat on their own ports seems a feasible solution.
From experience, I dislike the idea of going back to Apache+Tomcat.
Both are fine by themselves, but the combi is a hassle.
Any hints for better alternatives to reduce the impact of stuck servlets?
-- Dr. Oskar Bartenstein osk...@ifcomputer.co.jp IF Computer Japan http://www.ifcomputer.com