As a follow up to this thread, we believe we have found the culprits (2). 1. mod_jk has a defect where it mixes the request/response pairs and sends a response that does not correspond to the request.
https://issues.apache.org/bugzilla/show_bug.cgi?id=47714 <https://issues.apache.org/bugzilla/show_bug.cgi?id=47714>To work around the issue, we had to update the workers.properties and the Apache mod_jk directives to disable retries and disable socket re-use. Apache Directives: JkOptions +DisableReuse worker.properties: worker.propertiese.retries=0 2. The header size periodically exceeded the max packet size supported by the AJP configuration. When the headers do not fit into a single packet, AJP drops the packet and logs an error. We resolved this by increasing the max packet size to 65536. This is more of an issue where special things are inserted into the request/response headers (SSL, SSO, other things). To address this we had to update the Tomcat AJP listener configuration (server.xml) and the worker.properties: worker.properties: worker.midtier.max_packet_size=65536 Tomcat AJP Listener (added the packetSize parameter): <Connector port="8009" protocol="AJP/1.3" redirectPort="8443" packetSize="65536"/> Axton Grams The opinions, statements, and/or suggested courses of action expressed in this E-mail do not necessarily reflect those of BMC Software, Inc. My voluntary participation in this forum is not intended to convey a role as a spokesperson, liaison or public relations representative for BMC Software, Inc. On Fri, Jul 30, 2010 at 2:09 PM, cpgold <[email protected]> wrote: > ** > We see these errors on 5.5.20 Tomcat (Solaris 9) on QST. A full restart of > Tomcat was required. > I haven't seen these errors on Development with Tomcat 6.0.28 (Solaris 9) > > --Chuck > > On Tue, Jul 27, 2010 at 10:42 AM, Axton <[email protected]> wrote: > >> ** Unfortunately, bypassing Apache altogether is not an option as our SSO >> authentication is handled through an Apache plug-in. >> >> Axton Grams >> >> The opinions, statements, and/or suggested courses of action expressed in >> this E-mail do not necessarily reflect those of BMC Software, Inc. My >> voluntary participation in this forum is not intended to convey a role as a >> spokesperson, liaison or public relations representative for BMC Software, >> Inc. >> On Tue, Jul 27, 2010 at 9:44 AM, LJ LongWing <[email protected]>wrote: >> >>> ** >>> >>> Axton, >>> >>> Our configs are surprisingly similar….we are using the WebLogic server as >>> both Web and JSP at this point…we don’t seem to be having issues with the >>> WebLogic with no Proxy in place…nor do we seem to have issues when using the >>> Tomcat that comes with 7.5 (5.5.28)…so I suggest trying to bypass Apache all >>> together and see if that eliminates your issue…just a thought. >>> >>> >>> >>> Mid Tier Version >>> >>> 7.5.00 Patch 005 201005111542 >>> >>> Installation Directory >>> >>> /app/bmc/midtier >>> >>> Web Server Information >>> >>> WebLogic Server 10.3.2.0 Tue Oct 20 12:16:15 PDT 2009 1267925 Oracle >>> WebLogic Server Module Dependencies 10.3 Tue Oct 20 13:57:01 EDT 2009 >>> >>> Operating System Name >>> >>> SunOS >>> >>> Java Version >>> >>> 1.6.0_18 >>> >>> >>> >>> >>> >>> *From:* Action Request System discussion list(ARSList) [mailto: >>> [email protected]] *On Behalf Of *Axton >>> *Sent:* Monday, July 26, 2010 8:36 PM >>> *To:* [email protected] >>> *Subject:* Re: Mid-Tier 7.5 Problems >>> >>> >>> >>> ** That's the same situation I am in. Getting logs is troublesome, at >>> best. Reproducing the errors seems to be a draw of the hat (maybe 1 in 30 >>> chance), so we too are having a near impossible time getting logs of the >>> errors. The errors seem to persist with a given session; meaning that I can >>> consistently reproduce the errors with a given set of actions when I get a >>> bad session, but as soon as I log out and log back in, the errors stop in >>> that given scenario. It's as if something associated with the session >>> becomes corrupted; it doesn't seem to be anything that is cached on the >>> client side, as clearing the cache while the error happens does not resolve >>> the issue for that session; but creating a new session does. >>> >>> >>> >>> What version of Java were you using when you ran into the issues. We are >>> using the following: >>> >>> >>> >>> Tomcat 6.0.20 >>> >>> Sun Java 1.6.0_17 >>> >>> Apache 2.0.63 >>> >>> mod_jk 1.2.28 (we were using 1.2.30, but had more problems with that >>> version) >>> >>> Solaris 9 >>> >>> Mid-Tier 7.5 patch 4 >>> >>> >>> >>> We are getting ready to try mod_proxy instead of mod_jk to see if this >>> resolves the issues. >>> >>> >>> >>> We are seeing the errors most around actions that pull table results as >>> well. >>> >>> >>> >>> Let me know if your issues stop after your move to WebLogic. >>> >>> >>> >>> Axton >>> >>> >>> >>> The opinions, statements, and/or suggested courses of action expressed in >>> this E-mail do not necessarily reflect those of BMC Software, Inc. My >>> voluntary participation in this forum is not intended to convey a role as a >>> spokesperson, liaison or public relations representative for BMC Software, >>> Inc. >>> >>> >>> >>> >>> >>> On Mon, Jul 26, 2010 at 5:01 PM, LJ LongWing <[email protected]> >>> wrote: >>> >>> ** >>> >>> Axton, >>> >>> We are seeing lots of ‘Caught Exception’ errors….we have been struggling >>> to get them fixed even with vendor support…I have also received a single ‘if >>> you are on SSO’ report, but only one. >>> >>> >>> >>> I haven’t investigated the SSO error because I have only gotten it once >>> >>> >>> >>> The ‘Caught Exception’ errors….we are switching from Tomcat to WebLogic >>> servers, and upgrading from one version of a SunOne proxy server to >>> another….too many changes to be able to track which one is the cause…we have >>> found that the problem doesn’t seem to exist when hitting the web servers >>> directly/bypassing the proxy….we have been able to narrow it down to what >>> seems to be URL length issues making backchannel calls. This seems to >>> happen most often when workflow is doing a refresh of a table field, but not >>> able to get it to happen consistently enough to get good logs to send to >>> BMC…. >>> >>> >>> >>> Let me know if any of this is helpful/if you have anything to add to help >>> us fix… >>> >>> >>> >>> *From:* Action Request System discussion list(ARSList) [mailto: >>> [email protected]] *On Behalf Of *Axton >>> *Sent:* Monday, July 26, 2010 3:24 PM >>> *To:* [email protected] >>> *Subject:* Mid-Tier 7.5 Problems >>> >>> >>> >>> ** Wanted to see if anyone out there is having these types of issues with >>> Mid-Tier 7.5 patch 4 or higher: >>> >>> - blank page when accessing a form >>> >>> - page stops loading at the animated gif (loading) >>> >>> - Javascript 'Caught Exception' errors >>> >>> - "Unable to reach Mid-Tier. If you are on SSO, please re-login to SSO >>> service." errors >>> >>> _attend WWRUG10 www.wwrug.com ARSlist: "Where the Answers Are"_ >>> >>> >>> >>> >>> >>> _attend WWRUG10 www.wwrug.com ARSlist: "Where the Answers Are"_ >>> _attend WWRUG10 www.wwrug.com ARSlist: "Where the Answers Are"_ >>> >> >> _attend WWRUG10 www.wwrug.com ARSlist: "Where the Answers Are"_ >> > > _attend WWRUG10 www.wwrug.com ARSlist: "Where the Answers Are"_ > _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at www.arslist.org attend wwrug10 www.wwrug.com ARSlist: "Where the Answers Are"
