tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Konstantin Preißer <kpreis...@apache.org>
Subject RE: svn commit: r1530081 - /tomcat/trunk/java/org/apache/tomcat/util/net/AprEndpoint.java
Date Wed, 09 Oct 2013 01:41:20 GMT
Mark,

> > -----Original Message-----
> > From: Mark Thomas [mailto:markt@apache.org]
> > Sent: Tuesday, October 8, 2013 3:57 PM
> > To: Tomcat Developers List
> > Subject: Re: svn commit: r1530081 -
> > /tomcat/trunk/java/org/apache/tomcat/util/net/AprEndpoint.java
> >
> >
> > None. I've seem the same problem but haven't got to the bottom of it
> > yet. What I do know is:
> > - Tomcat doesn't send a close
> > - The browser closes the connection on its own
> > - Tomcat tries (and fails) to write some data and that triggers the
> >   close on the Tomcat side
> >
> > I'm wondering if some form of data corruption could be the trigger.
> >
> > Whatever is going on, I'm fairly sure it is a different issue to BZ 55633.
> >
> > I need to do some more careful analysis with Wireshark.
> 
> OK, thanks.
> 
> Note: When I tested this scenario, I got a slightly other result: For me, Tomcat
> initiated the shutdown of TCP connection. (And another issue: When/After
> this happens, if the browser tries to establishes a new Websocket
> connection (with a Upgrade: ... header) then tomcat sometimes shuts down
> the TCP connection without sending a reply).

I do now see the behavior which you described, but with the NIO connector: There the browser
is the one which closes the connection (probably due to a data corruption).

E.g. when redirecting IE and make it stop to read incoming data for ~ 30 seconds, then make
it to continue reading, IE will display (translated from german): "SCRIPT12152: WebSocket
Error: Network Error 12152, The server sent an invalid or unknown reply" and reset the TCP
connection.

When I tried this with FF, it first seemed that here Tomcat was the one which initiated a
normal TCP shutdown, but on more debugging it seems that Firefox stopped at some point to
process the received messages and then sent something like a Websocket Close message so that
Tomcat closes the connection (when pausing the Firefox -> Tomcat connection prior to unpausing
Tomcat -> Firefox direction, Tomcat continues to send data).

When trying with APR, I keep seeing that Tomcat initates the connection shutdown.

With BIO, the problems don't occur.


Hope this helps,


Regards,
Konstantin Preißer


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org


Mime
View raw message