Eric
This is because the ESB tries to use keepalive connections as much as
possible for obvious performance reasons, and I do not want to recommend
falling back to HTTP 1.0 unless the other party does not understand HTTP
1.1..
From my post
<http://mail-archives.apache.org/mod_mbox/hc-dev/200805.mbox/[EMAIL PROTECTED]>
to the HttpCore list last night, this is something we need to expect and
handle gracefully, and JBoss (or whoever) is free to do whatever they
choose to ensure its proper functioning under load.
I have the fix for this issue right now, and it just needs a bit more
testing and verification, esp to ensure that SYNAPSE-255 remains fixed
as well, which I will check with Ruwan once he gets to work today.
Thanks for all your help and patience in helping us nail this issue, and
I hope you feel better now.. esp with this news :-)
thanks
asankha
Hubert, Eric wrote:
Hi Sanjiva,
indeed this could be a valid approach, but I have to add some details. If Asankha is right with his evaluation, and I'm quite sure he is, then this is an issue which on the one hand needs to be fixed properly but on the other hand does not block us at all.
Unfortunately we currently don't use keep alive at all. Since a few days I know why. The developers of our client-side dynamic proxy had to deactivate keep alive due to a bug in the some bea web bridge. I don't know the details. My hope was that we can change this after the migration from Bea to JBoss has been finished. For now we might have to live without keep alive anyhow. :-(
I just stumbled about that problem as I tried to test the graceful instance shutdown, which we really need - clusterwide. Beause I didn't know the cause for the issue, it worried me a lot.
Regards,
Eric
I'll re-run the same test without having keep alive activated in SoapUI just to
verify our current understand is correct. Will get back to you soon.
_______________________________________________
Esb-java-dev mailing list
[email protected]
http://wso2.org/cgi-bin/mailman/listinfo/esb-java-dev
