Salut,
back on the radar :-)
Kristian Rink wrote:
> Salut Jeanfrancois, *;
>
> Kristian Rink schrieb:
>
>> For what I see, the default behaviour is sending it out on the fly but I
>> will check back on that the next days.
>
> As stated, some more (final?) words on this issue: So far, I have carefully
> configured our apache2 to disable any sort of KeepAlive wherever possible -
> global KeepAlive off, forcing no-keepalive and "old" HTTP 1.0 for proxy
> request in the mod_proxy configuration...
>
> SetEnv force-proxy-request-1.0 1
> SetEnv proxy-nokeepalive 1
> SetEnv proxy-sendchunked 1
>
>
> ..., disabling keep-alive in the glassfish backend server. Result:
>
> - The number of ClientAbortExceptions reported has drastically decreased,
> though not completely gone. Same has the amount of people complaining about
> this issue - even though I recieve one of the screenshots reporting "site
> not found" or "site not responding" once in a while, it is nowhere next to
> where we started a while ago which is good (even though the number of these
> reports being zero would be not that bad, as well... :) ).
>
> - I do have a reliable feedback from at least one customer who reported
> that, after turning off keep-alive in our configuration, the download using
> MSIE did work again.
>
> So by now I consider this to be the reason for our pain and gonna keep a
> note in our documentation system to pay attention to that whenever deploying
> a new proxy system again. Apart from that, are there any other things that
> should be learnt from this? Is this behaviour just the consequence of an
> annoying bug or does something need to be done about this?
>
Can you share a test case/configuration I can play with it. The MSIE
problem is something interesting as the NIO implementation with the JDK
doesn't react the same way when MSIE is used.
A+
-- Jeanfrancois
> Anyway, thanks a bunch for your enduring support getting me/us through this,
> much appreciated. :)
>
> Cheers & kind regards,
> Kristian
>