Re: wc_get_worker_for_name, done did not found a worker - Plea for
help 2004-06-01 - By Barry Roberts
Back
Oops, I just noticed that I cut and pasted from two different places.
My httpd.conf from apache1 is looking for a worker lb and the
workers.properties from apache2 doesn't have that. Here are the correct
lines from the apache2 httpd.conf
JkWorkersFile /usr/local/apache2/conf/workers.properties
JkLogFile /var/log/mod_jk.log
JkLogLevel debug
JkMount /apps/* xsappa
JkMount /admin xsappa
On Tue, 2004-06-01 at 09:37, Barry Roberts wrote:
> I'm getting desperate to get mod_jk working in apache on RHES 3 talking
> to tomcat 5.0.19 on (another) RHES3. I'm getting the error from the
> subject line if I set JkLogLevel to "debug." In the browser I see 500's
> for ANY servlet.
>
> I've tried the RHES3 apache2 rpms (httpd-2 (See http://tpd-2.ora-code.com).0.46-25.ent), building
> apache2 from source (2.0.49), building apache1 from source (1.3.31),
> with mod_jk static or DSO. Always the same thing.
>
> I've played with the permissions on the workers.properties to world
> readable, writable, and even executable. I strace'd httpd and it
> appears to successfully read the file (strace output below).
>
> I even copied a known workers.properties from a production web server
> (apache 2.0.43 built from source on Solaris) and it does the same
> thing. So I'm reasonably sure it's not a problem with the
> workers.properties file itself.
>
> I built mod_jk from the
> jakarta-tomcat-connectors-jk-1.2-src-current.tar.gz tarball which
> creates a jakarta-tomcat-connectors-jk-1.2.5-src directory.
>
> I built my own version of mod_jk that logs in the map_get function in
> jk_map.c and it appears that the map passed in the parameter jk_map_t *m
> is null. I've been trying to figure out why, and I'm just about stuck.
>
> I have also tried mod_jk2 (with a workers2.properties) and it works
> under low load. Even with only 3 thread in siege I get "Broken Pipe"
> errors in the tomcat logs and AjpGetReply errors in the jk logs. I
> think this is similar to bugs 28790 and/or 28532. We also saw this in
> our production servers, and our customers were seeing pages that just
> never finished loading until we switched back to jk1. So I could use
> jk2 if I knew how to get around that, but jk1 would do, too.
>
> Relevant (I think) strace output:
>
> stat64("/usr/local/apache/conf/workers.properties",
> {st_mode=S_IFREG|0777, st_size=788, ...}) = 0
> read(3, "", 4096) = 0
> close(3) = 0
> munmap(0xb75ea000, 4096) = 0
>
> from httpd.conf:
>
> JkWorkersFile /usr/local/apache/conf/workers.properties
> JkLogFile /var/log/mod_jk-a13.log
> JkLogLevel debug
> JkMount /apps/* lb
>
> workers.properties:
> worker.list=xsappa
>
> worker.xsappa.port=139
> worker.xsappa.host=192.168.2.40
> worker.xsappa.type=ajp13
> worker.xsappa.lbfactor=1
>
>
> -- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ------
> To unsubscribe, e-mail: tomcat-user-unsubscribe@(protected)
> For additional commands, e-mail: tomcat-user-help@(protected)
>
-- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ------
To unsubscribe, e-mail: tomcat-user-unsubscribe@(protected)
For additional commands, e-mail: tomcat-user-help@(protected)
|
|
