PQ72326: WHEN KEEPALIVE IS EXCEEDED AND SHUTS DOWN INTERNAL CONNECTIONS IT IS PROPAGATING OUT TO THE EXTERNAL CLIENT WEBSERVER | |||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||||||||||||||||||||
![]() APAR status Closed as program error. Error description to not be closed by WAS. Instead it should just be shutting down the associated KeepAlive connection between the internal HTTP Transport of WAS and the external IHS. Example: % TestXML -x XML/pslping.xml -s /XML_TCPP_T -p 11202 -h mauiwdmmk2 -l 5HOST = mauiwdmmk2 Start TestXML at: 12:46:43.514 Post to URL : /XML_TCPP_T HTTP/1.1 Host:mauiwdmmk2:11202 Http Connection Type : close ===== LOOP 1 ===== Test Case : XML/pslping.xml XML Send Size : 232 XML Received Size : 430 +------ Dump Reply -----+ | | Http Version : HTTP/1.1 | Status : 200 | Reason : OK | +------ Entity Dump -----+ | | [Date]=[Thu, 14 Nov 2002 17:46:43 GMT] | [Server]=[IBM_HTTP_SERVER/1.3.19.2 Apache/1.3.20 (Unix)] | [Content-Length]=[430] | [Connection]=[close] | [Content-Type]=[text/xml] | [Content-Language]=[en] +------------------------+Local fix Debug module provided by L3 for testing. They want to see if this module behaves more like the behavior you are expecting. If it does, then an APAR will be created and an official efix developed. 1) Stop the webserver. 2) Backup the existing mod_ibm_app_server_http.so in the WebSphere/AppServer/bin directory. 3) Copy the attached file into the WebSphere bin directory. 4) Start the webserver. 5) Run the test. Customer indicated this had no effect on the problem but asked why their httpd.conf file was pointing to mod_app_server_http.so and the module provided and installed in WAS was mod_ibm_app_server_http.so. Turns out they had httpd.conf file pointing to the Apache module instead of the IBM IHS module. Had customer cahnge the httpd.conf file to point to th ecorrect module. Still no effect. L3 suggested one last change, add a VirtualHost alias in the Host Aliases of WebSphere for <hostname>:80 and regen the plugin. Once this was done, the errors were eliminated and the KeepAlive functioned as desired.Problem summary **************************************************************** * USERS AFFECTED: WebSphere Application Server version 4.0 * * users of the webserver plugins. * **************************************************************** * PROBLEM DESCRIPTION: The plugin would send the connection * * header received from the app server * * response back to the client. This * * could result in connections closing * * that could have been Keep-Alive * * connections. * **************************************************************** * RECOMMENDATION: * **************************************************************** The plugin should not propagate the Connection header from the app server response in the response to the real client. The webserver should be in charge of whether that connection should be persistent or not.Problem conclusion The plugin now removes the connection header from the app server response and allows the webserver to handle whether the connection with the client is persistent or not.Temporary fix The fix has been delivered to the customer and we are awaiting verification it is successful.Comments
APAR is sysrouted FROM one or more of the following: PQ68477 APAR is sysrouted TO one or more of the following: Modules/Macros
Publications Referenced
|
Product categories: Software > Application Servers >
Distributed Application & Web Servers > WebSphere Application
Server > General
Operating system(s):
Software version: 00S
Software edition:
Reference #: PQ72326
IBM Group: Software Group
Modified date: Mar 21, 2003
(C) Copyright IBM Corporation 2000, 2008. All Rights Reserved.