> Thank you for the suggestion.
> Did you file a bug report to the glassfish team, and
> is it publicly available?
> I'd be more comfortable with an "official" evaluation
> of the problem and a supported
> workaround/correction.
We did open a case through our SUN support contact. So it is not publicly available.
Unfortunately we have no workaround nor fix proposed by SUN on this one.
They only suggest us to test with woodstox with the results above.
Note also that we are able to reproduce with very low load (2 concurrent users) ...
I have also noticed those two bugs:
https://glassfish.dev.java.net/issues/show_bug.cgi?id=8518
Issue #|8518
Summary|Problem with SOAP messages with dependancy on the HTTP header length
Component|glassfish
Version|v2.1.1
https://glassfish.dev.java.net/issues/show_bug.cgi?id=8562
Issue #|8562
Summary|POST request with double content-length header should throw bad request error
Component|glassfish
Version|V3
Those two bugs describe behaviours that we have observed during our load tests.
I think we could talk about corrupted SOAP responses in these cases also.
The issue 8562 suggests that a test has been made over v2.1.
Unfortunately we also have a case opened about GlassFish issueing requests containing two content-length header whereas our web services does provide only one (cross checked with the use of other well known application servers).
Best Regards
[Message sent by forum member 'vkoniecz' (vkoniecz)]
http://forums.java.net/jive/thread.jspa?messageID=352474