Hello Tomcat Users and Committers, [duplicate as first msg got formatted
incorrectly]Platform: Tomcat 4.1.24, Linux ES 2.1, jdk1.4.1.06Here is a snippet
of the stack trace we regularly in our Tomcat 4.1.24 logs.----- Root Cause
-----java.lang.IllegalStateException: getOutputStream() has already been called
for this response at
org.apache.catalina.connector.ResponseBase.getWriter(ResponseBase.java:756)
at
org.apache.catalina.connector.ResponseFacade.getWriter(ResponseFacade.java:174)
at
org.apache.jasper.runtime.JspWriterImpl.initOut(JspWriterImpl.java:173)
at org.apache.jasper.runtime.JspWriterImpl.flushBuffer(JspWriterImpl.java:166)
at
org.apache.jasper.runtime.PageContextImpl.release(PageContextImpl.java:184)
at
org.apache.jasper.runtime.JspFactoryImpl.internalReleasePageContext(JspFactoryImpl.java:198)
at
org.apache.jasper.runtime.JspFactoryImpl.releasePageContext(JspFactoryImpl.java:193)
at
org.apache.jsp.ViewAttachment_jsp._jspService(ViewAttachment_jsp.java:76)We
intend to migrate to Tomcat 5.x later this year. In the interim, we recently
linked this exception event to leaked memory. That is, the
PageContextImpl.release() method (6th from top in stack trace above, body
below) never completes as it does not catch the IllegalStateException that
happens when ResponseBase.getWriter()is called in the scope of 2nd
flushBuffer() call below.See org.apache.jasper.runtime.PageContextImpl:public
void release() { out = baseOut; try { if (isIncluded) {
((JspWriterImpl)out).flushBuffer(); // push it into the including
jspWriter } else { // Do not flush the buffer even if we're
not included (i.e. // we are the main page. The servlet will flush
it and close // the stream.
((JspWriterImpl)out).flushBuffer(); }} catch (IOException ex) {
loghelper.log("Internal error flushing the buffer in release()");}servlet
= null;config = null;context = null;needsSession =
false;errorPageURL = null;bufferSize = JspWriter.DEFAULT_BUFFER;autoFlush
= true;request = null;response = null; depth =
-1;baseOut.recycle();session = null;attributes.clear();}The 2nd last line
above, session=null is skipped because of the uncaught exception, which creates
a problem since a session so referenced becomes ineligible for garbage
collection after processing the request.Does anyone know:a) is this a
documented limitation (I read 102 matches for ‘memory leak’ on the Tomcat
4.1 list archives, but I did not find this exact problem, except for an entry
in the 4.1 release notes where it is reported fixed in 4.1.20).b) is there a
patch available to fix this in Tomcat 4.x?Thanks very much for your time,Matthew
