You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tomcat.apache.org by Mladen Turk <mt...@apache.org> on 2004/03/01 11:51:58 UTC
What's left for JK2.0.4?
Henri and others,
What are the remaining issues to be solved?
How about inproc JNI?
I'd like to disable that for all multi process mpm's, meaning almost
everything except winnt.
(It doesnt't work or ever will).
What is the status of shm?
Again, it's useless for all single mpm's.
How about to disable that for those mpm's that don't need this?
MT.
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
jk2 / apr_socket
Posted by Henri Gomez <hg...@apache.org>.
Seems there is something wrong here, 3 params awaited but only
2 params provided:
if (remote_sa->next) {
env->l->jkLog(env, env->l, JK_LOG_DEBUG,
"channelApr.open(): error %d creating
socket %d %s\n",
ret, socketInfo->host);
}
else {
env->l->jkLog(env, env->l, JK_LOG_ERROR,
"channelApr.open(): error %d creating
socket %d %s\n",
ret, socketInfo->host);
}
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
inet_ntoa and multi-threaded OS/APPS
Posted by Henri Gomez <hg...@apache.org>.
In jk there is many areas where we need to add informations about
the remote tomcat IP adress and we only a sockaddr_in struct available.
Question:
There is the inet_ntoa function but I wonder how it works with
multi-threaded apps on OS like Windows, Netware, ...
Regards.
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
Re: What's left for JK2.0.4?
Posted by Henri Gomez <hg...@apache.org>.
Mladen Turk wrote:
>
>
>
>>-----Original Message-----
>>From: Henri Gomez
>>
>>>Yes, the ones that create multiple child processes.
>>>Each child in that case wishes to run it's own Tomcat process, with
>>>the same config.
>>
>>>The consecutive TC instance fails to load (same config, same ports).
>>
>>Correct indeed, how could we overcome this ?
>>
>
>
> Using TC in ebedded mode and making JNI wrapper for configuration.
> But that's for JK3 thought.
>
>
>
>>>I don't have access for a moment to any nix machine so
>>
>>couldn't test that.
>>
>>Should we delay this for jk 2.0.5 ?
>>
>
>
> I don't think so.
>
> I'll make the patches and try to simulate that on windows.
> (I have my own multi child winnt-mpm)
>
> But someone will have to test that on some nix with worker and prefork mpm.
I'll do the test on Linux with worker/prefork MPM
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
RE: What's left for JK2.0.4?
Posted by Mladen Turk <mt...@apache.org>.
> -----Original Message-----
> From: Henri Gomez
> >
> > Yes, the ones that create multiple child processes.
> > Each child in that case wishes to run it's own Tomcat process, with
> > the same config.
>
> > The consecutive TC instance fails to load (same config, same ports).
>
> Correct indeed, how could we overcome this ?
>
Using TC in ebedded mode and making JNI wrapper for configuration.
But that's for JK3 thought.
> >
> > I don't have access for a moment to any nix machine so
> couldn't test that.
>
> Should we delay this for jk 2.0.5 ?
>
I don't think so.
I'll make the patches and try to simulate that on windows.
(I have my own multi child winnt-mpm)
But someone will have to test that on some nix with worker and prefork mpm.
MT.
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
Re: What's left for JK2.0.4?
Posted by Henri Gomez <hg...@apache.org>.
Mladen Turk wrote:
>
>
>
>>-----Original Message-----
>>From: Henri Gomez
>>
>>>How about inproc JNI?
>>>I'd like to disable that for all multi process mpm's,
>>
>>meaning almost
>>
>>>everything except winnt.
>>>(It doesnt't work or ever will).
>>
>>Multi process MPM ? you means multi-threaded and prefork models ?
>>
>
>
> Yes, the ones that create multiple child processes.
> Each child in that case wishes to run it's own Tomcat process, with the same
> config.
> The consecutive TC instance fails to load (same config, same ports).
Correct indeed, how could we overcome this ?
>>>What is the status of shm?
>>>Again, it's useless for all single mpm's.
>>>How about to disable that for those mpm's that don't need this?
>>
>>It seems we also have a problem in Apache 2.0 at init time,
>>where I see many message telling that jk2 can't find child
>>XXX in the 256 scoreboard slots
>>
>
>
> IMO we should use the Apache scoreboard image.
> If it's not present then our shm shouldn't be too.
> also instead getpid() call we should use:
> ap_get_scoreboard_process(i); (Look at the mod_status)
I'll take a look at this.
> Then set our own worker->childId to be the scoreboard slot number not the
> process id.
>
> I don't have access for a moment to any nix machine so couldn't test that.
Should we delay this for jk 2.0.5 ?
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
RE: What's left for JK2.0.4?
Posted by Mladen Turk <mt...@apache.org>.
> -----Original Message-----
> From: Henri Gomez
> >
> > How about inproc JNI?
> > I'd like to disable that for all multi process mpm's,
> meaning almost
> > everything except winnt.
> > (It doesnt't work or ever will).
>
> Multi process MPM ? you means multi-threaded and prefork models ?
>
Yes, the ones that create multiple child processes.
Each child in that case wishes to run it's own Tomcat process, with the same
config.
The consecutive TC instance fails to load (same config, same ports).
> > What is the status of shm?
> > Again, it's useless for all single mpm's.
> > How about to disable that for those mpm's that don't need this?
>
> It seems we also have a problem in Apache 2.0 at init time,
> where I see many message telling that jk2 can't find child
> XXX in the 256 scoreboard slots
>
IMO we should use the Apache scoreboard image.
If it's not present then our shm shouldn't be too.
also instead getpid() call we should use:
ap_get_scoreboard_process(i); (Look at the mod_status)
Then set our own worker->childId to be the scoreboard slot number not the
process id.
I don't have access for a moment to any nix machine so couldn't test that.
MT.
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org
Re: What's left for JK2.0.4?
Posted by Henri Gomez <hg...@apache.org>.
Mladen Turk wrote:
> Henri and others,
>
> What are the remaining issues to be solved?
>
> How about inproc JNI?
> I'd like to disable that for all multi process mpm's, meaning almost
> everything except winnt.
> (It doesnt't work or ever will).
Multi process MPM ? you means multi-threaded and prefork models ?
> What is the status of shm?
> Again, it's useless for all single mpm's.
> How about to disable that for those mpm's that don't need this?
It seems we also have a problem in Apache 2.0 at init time, where
I see many message telling that jk2 can't find child XXX in the 256
scoreboard slots
---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-dev-help@jakarta.apache.org