|
Home > Archive > Netware Webserver > December 2007 > Potential Tomcat Issue - Service Temporarily Unavailable
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Potential Tomcat Issue - Service Temporarily Unavailable
|
|
|
| We're running Netware 6.5 SP6. At random, we'll receive "Service
Temporarily Unavailable" followed by "The server is temporarily unable to
service your request due to maintenance downtime or capacity problems.
Please try again later.
Apache/2.0.59 (NETWARE) mod_jk/1.2.15 Server at <server ip> Port 80" when
trying to access Webaccess.
By unloading Tomcat and Apache, we're good for another 5 minutes to 1 day.
This has become a huge problem. Most times I have to do a Java -killall
before I can get Tomcat to load properly after stopping it.
While this is happening, our Apache mod_jk.log displays "(929): sendfull
returned -53 with errno=53" and "(1758): Error connecting to tomcat.
Tomcat is probably not started or is listening on the wrong port.
worker=ajp13 failed".
I have tried doing a tckeygen on one of our servers and this problem still
persist. Apache and Tomcat are clustered on their on cluster resource. I
really appreciate any help anyone can provide. Thanks.
| |
| Guenter 2007-11-23, 7:11 pm |
| Hi Ryan,
Ryan@Ryan.com (Ryan) wrote in news:oLD1j.544$k06.12@kovat.provo.novell.com:
> I really appreciate any help anyone can provide. Thanks.
you'r mod_jk is meanwhile 10 versions behind (not to talk about Tomcat);
so I'd suggest you save your existing mod_jk, and try latest from here:
http://www.apache.org/dist/tomcat/t...naries/netware/
please let me know how that goes....
Günter.
| |
|
| Thanks a lot for the tip. I have updated this module to 1.2.25. I will see
how it goes and report back. Should I also update Tomcat? Would my (NW6.5
SP6) server be able to run the latest release? Thanks again.
Ryan
Guenter wrote:
> you'r mod_jk is meanwhile 10 versions behind (not to talk about Tomcat);
> so I'd suggest you save your existing mod_jk, and try latest from here:
> http://www.apache.org/dist/tomcat/t...naries/netware/
> please let me know how that goes....
> Günter.
| |
| Guenter 2007-11-24, 1:11 pm |
| Hi Ryan,
Ryan@Ryan.com (Ryan) wrote in news:afV1j.677$k06.455@kovat.provo.novell.com:
> Thanks a lot for the tip. I have updated this module to 1.2.25. I will see
> how it goes and report back. Should I also update Tomcat? Would my (NW6.5
> SP6) server be able to run the latest release? Thanks again.
hmm, Tomcat isnt that easy to update as Apache2 and its modules...
I can say for sure that latest TC 4.1.x and 5.0.x will run - I've heard that
the 5.5.x line should also work, but I'd suggest you wait few days to see if
the issue appears again; also its possible that the issue is not JK / TC
related, but may be caused by the Winsock stack, so I would first look for a
ws2 / tcpip update before I would touch the TC.
The reason why I suggested the mod_jk update first is because I know we did a
bunch of code changes since 1.2.15, and some might be important to your
issue.
If the problem is still there it might help to set loglevel to debug for
getting more details about the failure.
Günter.
| |
|
| Thanks for all of your help Guenter. Our Tomcat issue was resolved after
performing the Winsock update. I actually have an incident opened with
Novell now. Our last nagging issue is that Apache needs to be restarted
every morning due to slowdown issues. Webaccess is slow in the morning
until Apache gets reloaded. It is then fast for the entire day without any
issues. We are only reloading Apache, nothing else.
The engineer I'm working with wants us to load Apache from SYS: rather
then a clustered volume so that will likely be the next step. We're
loading it into a protected memory space and we don't receive any startup
issues or errors when the slowdown occurs. If you have any thoughts I'd
appreciate hearing them. Otherwise, we're sitting much better now then we
were a short time ago. Thanks again.
Ryan
| |
| Guenter 2007-12-05, 1:11 pm |
| Hi Ryan,
Ryan@Ryan.com (Ryan) wrote in news:n1B5j.5396$k06.1153
@kovat.provo.novell.com:
> Thanks for all of your help Guenter. Our Tomcat issue was resolved after
> performing the Winsock update. I actually have an incident opened with
great!
> Novell now. Our last nagging issue is that Apache needs to be restarted
> every morning due to slowdown issues. Webaccess is slow in the morning
> until Apache gets reloaded. It is then fast for the entire day without
any
> issues. We are only reloading Apache, nothing else.
>
> The engineer I'm working with wants us to load Apache from SYS: rather
> then a clustered volume so that will likely be the next step. We're
I doubt that this will make any difference...
> loading it into a protected memory space and we don't receive any startup
> issues or errors when the slowdown occurs. If you have any thoughts I'd
> appreciate hearing them. Otherwise, we're sitting much better now then we
> were a short time ago. Thanks again.
yes, few I have....
1.) when Apache turns to be slow again then check witrh NRM how much memory
is allocated to Apache / Tomcat etc., and compare to the values you can see
after a fresh restart. It may turn out to be a memory problem, or something
heavily leaks.
2.) If you have a default configuration then you have probably rotated
logging enabled; if possible I would at least for a test disable that and
just write to one big logfile...
Also on my site you can find an Apache module which does the log rotate
which might work better than the piped logging stuff.
3.) Meanwhile Apache 2.2.x is our mainstream, and if you use the Apache
instance only for TC access then you can quickly do a test with Apache
2.2.6 which is recent version:
http://www.gknw.net/phpbb/viewtopic.php?t=402
the '+mods' archive contains already mod_jk - so no need for a separate
download; and since this archive extracts to /apache22 folder you can very
easily test without changing anything else; just stop your apache2, and
launch apache22 instead - of course after you have configured properly;
please transfer your settings from your apache2 httod.conf, dont just copy
over the httpd.conf, its incompatible! You can find a separate mod_jk.conf
in the conf/extra folder which you can include into the httpd.conf - look
at the end of the 2.2.x httpd.conf where I have added a couple of includes
but commented. Take also a look here:
http://www.gknw.net/phpbb/viewtopic.php?t=426
4.) If you use PHP5 in addition this might as well be the culprit - let me
know if that's the case, and I will come up with some more suggestions.
good luck with finding the remaining issue!
Günter.
| |
|
| Thanks for all of your help Guenter. Updating to the newest version of
Apache from the links that you provided has resolved this issue. I am just
thrilled to have these problems put to rest. Webacccess no longer grinds
to a halt over time and there hasn't been a need to reload Apache since we
implemented this newer version on the 6th. You have been a HUGE help and
this saves me a lot of hassles of doing what our assigned engineer
recommended (getting Apache to load and run from the SYS volume). On top
of that, it all would've been done for nothing. It almost certainly would
not have resolved our issue.
The email address I left is legit, what's before -OMMIT- then sub @ for
AT. If you can send me a mailing address I would like to show my
appreciation by sending you a gift certificate to get out and have supper
and a few drinks or something. Maybe include where you're from so I can
reference what's available in your area. Thanks again.
Ryan
| |
| Guenter 2007-12-10, 1:11 pm |
| Hello Ryan,
RyanReuterATsio.midco.netOMMIT@s.com (Ryan) wrote in news:Opd7j.7281
$k06.5480@kovat.provo.novell.com:
> Thanks for all of your help Guenter. Updating to the newest version of
your're welcome!
> Apache from the links that you provided has resolved this issue. I am
just
> thrilled to have these problems put to rest. Webacccess no longer grinds
> to a halt over time and there hasn't been a need to reload Apache since
we
> implemented this newer version on the 6th. You have been a HUGE help and
> this saves me a lot of hassles of doing what our assigned engineer
> recommended (getting Apache to load and run from the SYS volume). On top
> of that, it all would've been done for nothing. It almost certainly would
> not have resolved our issue.
good to hear;
we're soon going to release Apache 2.2.7, and there we have again a bunch
of fixes in; there's also still a memory leak issue, but I cant tell for
sure if the fix will make it into 2.2.7, but certainly into 2.2.8; so watch
out for the next updates - these new versions are not only just for showing
a new version number!
> The email address I left is legit, what's before -OMMIT- then sub @ for
> AT. If you can send me a mailing address I would like to show my
> appreciation by sending you a gift certificate to get out and have supper
> and a few drinks or something. Maybe include where you're from so I can
> reference what's available in your area. Thanks again.
well, I'm from Germany; if you want to donate then probably easiest is if
you look at my site's forum http://www.gknw.net/phpbb and use the PayPal
link in the upper right corner....
If you register with my forum (yeah, we've just discussed that the Captcha
is die-hard, but dont give up - just try a few times please) then you can
join groups there which get notified when new releases of Apache2 and
mod_jk are available (I know anyway first since I'm with the Apache and
mod_jk projects, and the mod_jk binaries are created by me)...
greets, Günter.
| |
|
| Well shoot. As soon as I post and think this issue is behind us, we start
noticing significantly degraded performance after about 4 days. I'm unable
to restart Apache right at this very moment but I can do so early in the
morning. I guess on the positive, it has been better.
Webaccess is currently still usable, it has just lost a lot of it's snap.
It's my best guess that after an Apache restart in the morning, it'll be
back to lightening fast speeds. Well I'm sorry to be that guy that just
won't go away but what do you think Guenter?
I'm not real sure how I check the memory that Apache is using. Tomcat's
memory usage looks fine. Do you think that perhaps it's a memory leak
issue that may be resolved by an up and coming release? As usual, I
appreciate your help and any pointers that you may have.
Thanks,
Ryan
| |
|
| Just a note that Apache got so slow I had to restart it during production.
We are back to razor sharp speeds with Webaccess. I just wanted to inform
you that a restart does increase our Webaccess speed. Thanks.
> Well shoot. As soon as I post and think this issue is behind us, we start
> noticing significantly degraded performance after about 4 days. I'm unable
> to restart Apache right at this very moment but I can do so early in the
> morning. I guess on the positive, it has been better.
> Webaccess is currently still usable, it has just lost a lot of it's snap.
> It's my best guess that after an Apache restart in the morning, it'll be
> back to lightening fast speeds. Well I'm sorry to be that guy that just
> won't go away but what do you think Guenter?
> I'm not real sure how I check the memory that Apache is using. Tomcat's
> memory usage looks fine. Do you think that perhaps it's a memory leak
> issue that may be resolved by an up and coming release? As usual, I
> appreciate your help and any pointers that you may have.
> Thanks,
> Ryan
| |
| Anders Gustafsson 2007-12-11, 7:12 am |
| Ryan,
> I'm not real sure how I check the memory that Apache is using.
>
Try seg.nlm
- Anders Gustafsson (Sysop)
The Aaland Islands (N60 E20)
Discover the new Novell forums at http://forums.novell.com
Novell does not monitor these forums officially.
Enhancement requests for all Novell products may be made at
http://support.novell.com/enhancement
| |
| Guenter 2007-12-12, 7:12 pm |
| Hi Ryan,
Ryan@Ryan.com (Ryan) wrote in
news:Efk7j.7558$k06.2836@kovat.provo.novell.com:
[vbcol=seagreen]
well, I cant say if your issues are caused by a memory leak, but I can tell
you that the next Apache 2.2.7 release will contain 2 memory relevant fixes
since the backports were just applied.
As I said before check with NRM, or with seg.nlm as Anders pointed out:
http://www.novell.com/coolsolutions/tools/14445.html
seg.nlm is especially useful since it writes to periodically to files, so you
should be able to track down if memory gets short when webacc gets slow...
Another thing you should check is all log files; probably Apache logs
warnings that its going out of threads or such, whatever....
also a reminder: have you disabled rotating logs?
or at least replaced by mod_log_rotate:
http://www.gknw.net/development/apache/httpd-
2.2/netware/modules/mod_log_rotate-2.2.6-nw.zip
have you checked the Tomcat logs?
And another option which I'd suggest to test after updating to 2.2.7 might be
to use mod_proxy_ajp instead of mod_jk ...
Günter.
| |
| Guenter 2007-12-25, 1:11 pm |
| Hi Ryan,
Ryan@Ryan.com (Ryan) wrote in
news:Efk7j.7558$k06.2836@kovat.provo.novell.com:
[vbcol=seagreen]
we have just released mod_jk 1.2.26:
http://www.apache.org/dist/tomcat/t...ies/netware/jk-
1.2.26/
perhaps you might want to try this.
Günter.
|
|
|
|
|