You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tomcat.apache.org by Henri Gomez <he...@gmail.com> on 2007/04/17 17:17:08 UTC
mod_jk on i5/OS v5R4
Hi to all,
I'm still trying to adapt jk to i5/OS v5r4 and see the following in
mod_jk.log (debug mode)
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_uri_worker_map.c
(423): rule map size is 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [error] mod_jk.c (2701):
Initializing shm:/www/dapserver/logs/jk.shm.6589 errno=3025. Load
balancing workers will not function properly.
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] mod_jk.c (2712):
Setting default connection pool max size to 250
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_map.c (699):
Checking for references with prefix worker. with wildcard (recursion
1)
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (239):
creating worker ajp13
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (144):
about to create instance ajp13 of ajp13
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (157):
about to validate and init ajp13
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (1996):
worker ajp13 contact is 'localhost:8009'
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2126):
setting endpoint options:
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2129):
keepalive: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2133):
timeout: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2137):
buffer size: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2141):
pool timeout: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2145):
connect timeout: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2149):
reply timeout: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2153):
prepost timeout: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2157):
recovery options: 0
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2161):
retries: 2
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2165):
max packet size: 8192
[Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2033):
setting connection pool size to 250 with min 125
[Tue Apr 17 16:23:44 2007] [6589:0038] [info] mod_jk.c (2743):
mod_jk/1.2.22 initialized
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_uri_worker_map.c
(423): rule map size is 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [error] mod_jk.c (2701):
Initializing shm:/www/dapserver/logs/jk.shm.6589 errno=3025. Load
balancing workers will not function properly.
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2712):
Setting default connection pool max size to 250
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_map.c (699):
Checking for references with prefix worker. with wildcard (recursion
1)
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (239):
creating worker ajp13
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (144):
about to create instance ajp13 of ajp13
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (157):
about to validate and init ajp13
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (1996):
worker ajp13 contact is 'localhost:8009'
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2126):
setting endpoint options:
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2129):
keepalive: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2133):
timeout: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2137):
buffer size: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2141):
pool timeout: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2145):
connect timeout: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2149):
reply timeout: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2153):
prepost timeout: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2157):
recovery options: 0
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2161):
retries: 2
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2165):
max packet size: 8192
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2033):
setting connection pool size to 250 with min 125
[Tue Apr 17 16:23:45 2007] [6589:0038] [info] mod_jk.c (2743):
mod_jk/1.2.22 initialized
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2651):
Attached shm:/www/dapserver/logs/jk.shm.6589
[Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2661):
Initialized mod_jk/1.2.22
It's strange but it seems the initialisation is done twice. And at the
first time it failed to open some files, like shm file.
Is it a specific i5/OS case or general ?
For instance, JkWorkersFile can't be found at first try and as such
module can't be loaded, so I suspect we should relax the stat or be
able to determine (and skip) the first init phase on i5/OS.
Regards
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Henri Gomez <he...@gmail.com>.
> IIRC there were at some time similar probs with mod_ssl which also handles its own log files;
> if mod_ssl works on i5/OS probably you can find the trick there in the code...?
>
> just an idea...
Good idea but the SSL support on i5/OS is not mod_ssl (specific IBM code)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Guenter Knauf <fu...@apache.org>.
Hi Henri,
> Well I add to remove the JkWorkersFile directive to make mod_jk start
> and have some debug.
> I suspect the problem is that in the first stage of init the user
> profile on i5/OS is different than the real user profile in the second
> stage.
> This profile didn't have access to JkWorkersFile and so mod_jk fail to
> load the WorkersFile and report as error in first stage.
> Question :
> - How could we determine that we're in the first stage ?
> - Could 'relax' file access check for instance in the first stage to
> avoid file access problem on i5/OS ?
IIRC there were at some time similar probs with mod_ssl which also handles its own log files;
if mod_ssl works on i5/OS probably you can find the trick there in the code...?
just an idea...
Guenter.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: [PROPOSAL] add Tomcat version output to snoop.jsp sample
Posted by Henri Gomez <he...@gmail.com>.
+1
2007/4/18, Guenter Knauf <fu...@apache.org>:
> Hi all,
> since I often test with different Tomcat versions I would find it useful to see the TC version too with the snoop sample, f.e. something like that (or also at the end added):
>
> --- snoop.jsp.orig Wed Apr 18 18:35:12 2007
> +++ snoop.jsp Wed Apr 18 18:41:46 2007
> @@ -19,6 +19,8 @@
> <body bgcolor="white">
> <h1> Request Information </h1>
> <font size="4">
> +Server Software: <%= getServletContext().getServerInfo() %>
> +<br>
> JSP Request Method: <% out.print(util.HTMLFilter.filter(request.getMethod())); %>
> <br>
> Request URI: <%= request.getRequestURI() %>
>
> do you agree with this small add?
>
> thanks, Guenter.
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: dev-help@tomcat.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: [PROPOSAL] add Tomcat version output to snoop.jsp sample
Posted by Mark Thomas <ma...@apache.org>.
Guenter Knauf wrote:
> Hi all,
> since I often test with different Tomcat versions I would find it useful to see the TC version too with the snoop sample, f.e. something like that (or also at the end added):
-1. This will create a minor information disclosure vulnerability.
The same (actually more) information is available from:
- the manager (requires a password to access)
- catalina version (only available from the command line)
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
[PROPOSAL] add Tomcat version output to snoop.jsp sample
Posted by Guenter Knauf <fu...@apache.org>.
Hi all,
since I often test with different Tomcat versions I would find it useful to see the TC version too with the snoop sample, f.e. something like that (or also at the end added):
--- snoop.jsp.orig Wed Apr 18 18:35:12 2007
+++ snoop.jsp Wed Apr 18 18:41:46 2007
@@ -19,6 +19,8 @@
<body bgcolor="white">
<h1> Request Information </h1>
<font size="4">
+Server Software: <%= getServletContext().getServerInfo() %>
+<br>
JSP Request Method: <% out.print(util.HTMLFilter.filter(request.getMethod())); %>
<br>
Request URI: <%= request.getRequestURI() %>
do you agree with this small add?
thanks, Guenter.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Henri Gomez <he...@gmail.com>.
Well I add to remove the JkWorkersFile directive to make mod_jk start
and have some debug.
I suspect the problem is that in the first stage of init the user
profile on i5/OS is different than the real user profile in the second
stage.
This profile didn't have access to JkWorkersFile and so mod_jk fail to
load the WorkersFile and report as error in first stage.
Question :
- How could we determine that we're in the first stage ?
- Could 'relax' file access check for instance in the first stage to
avoid file access problem on i5/OS ?
Regards and thanks again
2007/4/17, William A. Rowe, Jr. <wr...@rowe-clan.net>:
> Rainer Jung wrote:
> > That's what I expect too, but in the log file Henri posted at the
> > beginning of this thread, both runs were done by the same process and
> > thread id. So at least if pids have a meaning on iSeries, both runs are
> > done by the same process. That's different from the *nix world, where
> > there is a switch to a new parent in between (as long as I remember
> > correct).
>
> Nope, that's correct. Both config passes happen on the single, main
> process and thread. THEN following the second pass, processes are
> forked off, and the threads created in each process. THAT is the
> unix world of httpd.
>
> On Windows, both passes happen in both the parent and child processes,
> the parent is actually does not pass its config via fork() to the
> children.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: dev-help@tomcat.apache.org
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Rainer Jung wrote:
> That's what I expect too, but in the log file Henri posted at the
> beginning of this thread, both runs were done by the same process and
> thread id. So at least if pids have a meaning on iSeries, both runs are
> done by the same process. That's different from the *nix world, where
> there is a switch to a new parent in between (as long as I remember
> correct).
Nope, that's correct. Both config passes happen on the single, main
process and thread. THEN following the second pass, processes are
forked off, and the threads created in each process. THAT is the
unix world of httpd.
On Windows, both passes happen in both the parent and child processes,
the parent is actually does not pass its config via fork() to the
children.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Rainer Jung <ra...@kippdata.de>.
That's what I expect too, but in the log file Henri posted at the
beginning of this thread, both runs were done by the same process and
thread id. So at least if pids have a meaning on iSeries, both runs are
done by the same process. That's different from the *nix world, where
there is a switch to a new parent in between (as long as I remember
correct).
William A. Rowe, Jr. wrote:
> Rainer Jung wrote:
>> Hi Henri,
>>
>> Henri Gomez wrote:
>>> 2007/4/17, Rainer Jung <ra...@kippdata.de>:
>>>> Apache always initializes twice (all platforms), so this part is usual.
>>>>
>>>> What seems special though, is that on *nix platforms, the two init runs
>>>> are done by different processes, the second process replaces the first
>>>> one. Your mod_jk log file shows the same pid for both runs. So this
>>>> might be special and lead to unexpected results, because now things are
>>>> really done twice.
>>> Well I5/OS is running in MPM mode so...
>> Yes, Apache 2.0 always uses MPMs, but on the platforms I know they never
>> initialize both times with the same process.
>
> The parent process always pre-loads the config, runs the full config
> hooks to test that the module can run, unloads everything and then will
> start "for real".
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by "William A. Rowe, Jr." <wr...@rowe-clan.net>.
Rainer Jung wrote:
> Hi Henri,
>
> Henri Gomez wrote:
>> 2007/4/17, Rainer Jung <ra...@kippdata.de>:
>>> Apache always initializes twice (all platforms), so this part is usual.
>>>
>>> What seems special though, is that on *nix platforms, the two init runs
>>> are done by different processes, the second process replaces the first
>>> one. Your mod_jk log file shows the same pid for both runs. So this
>>> might be special and lead to unexpected results, because now things are
>>> really done twice.
>>
>> Well I5/OS is running in MPM mode so...
>
> Yes, Apache 2.0 always uses MPMs, but on the platforms I know they never
> initialize both times with the same process.
The parent process always pre-loads the config, runs the full config
hooks to test that the module can run, unloads everything and then will
start "for real".
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Rainer Jung <ra...@kippdata.de>.
Hi Henri,
Henri Gomez wrote:
> 2007/4/17, Rainer Jung <ra...@kippdata.de>:
>> Apache always initializes twice (all platforms), so this part is usual.
>>
>> What seems special though, is that on *nix platforms, the two init runs
>> are done by different processes, the second process replaces the first
>> one. Your mod_jk log file shows the same pid for both runs. So this
>> might be special and lead to unexpected results, because now things are
>> really done twice.
>
> Well I5/OS is running in MPM mode so...
Yes, Apache 2.0 always uses MPMs, but on the platforms I know they never
initialize both times with the same process.
>> Concerning workers.properties: If your JkWorkersFile is configured, but
>> cannot be found you should get an error "Can't find the workers file
>> specified" or "Error in reading worker properties from " in your apache
>> error log. You didn't post your config, maybe there's a typo in it?
>
> No the file exist but it seems the first time, the module couldn't
> access the file system (I didn't know why and wait more informations
> from IBM)
What I tried to say: the posted log does not give any indication, that
mod_jk did not find a workers.properties. I just try to understand your
report, and there is no log line in your posted log, which has to do
with not finding a workers.properties.
>> What does not work in your case (judging from the log), is opening a
>> shared memory file, it throws errors during both init phases. What's
>> errno 3025 on iSeries? Does www/dapserver/logs/ exist and is writable?
>
> Errno 3025 is file/dir not found and as you can see, the second time
> it works so it's really something related with I5/OS implementation
> (argh)
Again: in the log part you posted, it does not work the first time and
it does not work the second time. I read errno 3025 both times.
>> From this log excerpt, i don't know, what you mean by "the module can't
>> be loaded"?
>
> Well the Apache instance exit with error message in its job log (i5/OS
> job journal) about error during post-configuration. And if the first
> time the i5/OS can't stat the workers.properties, I suspect it fail
> and error when trying to read it.
How do you know, that it can't stat the workers.properties?
Any details on the error in the job journal, anything in your standard
apache error log?
Regards,
Rainer
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Henri Gomez <he...@gmail.com>.
2007/4/17, Rainer Jung <ra...@kippdata.de>:
> Apache always initializes twice (all platforms), so this part is usual.
>
> What seems special though, is that on *nix platforms, the two init runs
> are done by different processes, the second process replaces the first
> one. Your mod_jk log file shows the same pid for both runs. So this
> might be special and lead to unexpected results, because now things are
> really done twice.
Well I5/OS is running in MPM mode so...
> Concerning workers.properties: If your JkWorkersFile is configured, but
> cannot be found you should get an error "Can't find the workers file
> specified" or "Error in reading worker properties from " in your apache
> error log. You didn't post your config, maybe there's a typo in it?
No the file exist but it seems the first time, the module couldn't
access the file system (I didn't know why and wait more informations
from IBM)
> What does not work in your case (judging from the log), is opening a
> shared memory file, it throws errors during both init phases. What's
> errno 3025 on iSeries? Does www/dapserver/logs/ exist and is writable?
Errno 3025 is file/dir not found and as you can see, the second time
it works so it's really something related with I5/OS implementation
(argh)
> I find a little strange, that your connections pool get autoconfigured
> to the size 250. So it looks like you are using a multi-threaded MPM
> (worker?) with threads per process 250.
Yes
> From this log excerpt, i don't know, what you mean by "the module can't
> be loaded"?
Well the Apache instance exit with error message in its job log (i5/OS
job journal) about error during post-configuration. And if the first
time the i5/OS can't stat the workers.properties, I suspect it fail
and error when trying to read it.
yes Apache 2.x implementation on i5/OS is a different beast than stock
Apache 2.x and without help from IBM labs it will be difficult to see
what could be done.
BTW, I started to work on the v5R4 UTF mode and will commit more tomorrow
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: mod_jk on i5/OS v5R4
Posted by Rainer Jung <ra...@kippdata.de>.
Apache always initializes twice (all platforms), so this part is usual.
What seems special though, is that on *nix platforms, the two init runs
are done by different processes, the second process replaces the first
one. Your mod_jk log file shows the same pid for both runs. So this
might be special and lead to unexpected results, because now things are
really done twice.
Concerning workers.properties: If your JkWorkersFile is configured, but
cannot be found you should get an error "Can't find the workers file
specified" or "Error in reading worker properties from " in your apache
error log. You didn't post your config, maybe there's a typo in it?
What does not work in your case (judging from the log), is opening a
shared memory file, it throws errors during both init phases. What's
errno 3025 on iSeries? Does www/dapserver/logs/ exist and is writable?
I find a little strange, that your connections pool get autoconfigured
to the size 250. So it looks like you are using a multi-threaded MPM
(worker?) with threads per process 250.
From this log excerpt, i don't know, what you mean by "the module can't
be loaded"?
Regards,
Rainer
Henri Gomez wrote:
> Hi to all,
>
> I'm still trying to adapt jk to i5/OS v5r4 and see the following in
> mod_jk.log (debug mode)
>
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_uri_worker_map.c
> (423): rule map size is 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [error] mod_jk.c (2701):
> Initializing shm:/www/dapserver/logs/jk.shm.6589 errno=3025. Load
> balancing workers will not function properly.
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] mod_jk.c (2712):
> Setting default connection pool max size to 250
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_map.c (699):
> Checking for references with prefix worker. with wildcard (recursion
> 1)
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (239):
> creating worker ajp13
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (144):
> about to create instance ajp13 of ajp13
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_worker.c (157):
> about to validate and init ajp13
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (1996):
> worker ajp13 contact is 'localhost:8009'
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2126):
> setting endpoint options:
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2129):
> keepalive: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2133):
> timeout: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2137):
> buffer size: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2141):
> pool timeout: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2145):
> connect timeout: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2149):
> reply timeout: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2153):
> prepost timeout: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2157):
> recovery options: 0
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2161):
> retries: 2
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2165):
> max packet size: 8192
> [Tue Apr 17 16:23:44 2007] [6589:0038] [debug] jk_ajp_common.c (2033):
> setting connection pool size to 250 with min 125
> [Tue Apr 17 16:23:44 2007] [6589:0038] [info] mod_jk.c (2743):
> mod_jk/1.2.22 initialized
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_uri_worker_map.c
> (423): rule map size is 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [error] mod_jk.c (2701):
> Initializing shm:/www/dapserver/logs/jk.shm.6589 errno=3025. Load
> balancing workers will not function properly.
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2712):
> Setting default connection pool max size to 250
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_map.c (699):
> Checking for references with prefix worker. with wildcard (recursion
> 1)
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (239):
> creating worker ajp13
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (144):
> about to create instance ajp13 of ajp13
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_worker.c (157):
> about to validate and init ajp13
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (1996):
> worker ajp13 contact is 'localhost:8009'
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2126):
> setting endpoint options:
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2129):
> keepalive: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2133):
> timeout: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2137):
> buffer size: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2141):
> pool timeout: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2145):
> connect timeout: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2149):
> reply timeout: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2153):
> prepost timeout: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2157):
> recovery options: 0
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2161):
> retries: 2
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2165):
> max packet size: 8192
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] jk_ajp_common.c (2033):
> setting connection pool size to 250 with min 125
> [Tue Apr 17 16:23:45 2007] [6589:0038] [info] mod_jk.c (2743):
> mod_jk/1.2.22 initialized
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2651):
> Attached shm:/www/dapserver/logs/jk.shm.6589
> [Tue Apr 17 16:23:45 2007] [6589:0038] [debug] mod_jk.c (2661):
> Initialized mod_jk/1.2.22
>
> It's strange but it seems the initialisation is done twice. And at the
> first time it failed to open some files, like shm file.
>
> Is it a specific i5/OS case or general ?
>
> For instance, JkWorkersFile can't be found at first try and as such
> module can't be loaded, so I suspect we should relax the stat or be
> able to determine (and skip) the first init phase on i5/OS.
>
> Regards
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org