ANNOUNCEMENT: Live Wireshark University & Allegro Packets online APAC Wireshark Training Session
April 17th, 2024 | 14:30-16:00 SGT (UTC+8) | Online

Wireshark-users: Re: [Wireshark-users] WSDL / XML support?

From: Guy Harris <guy@xxxxxxxxxxxx>
Date: Wed, 12 Sep 2007 15:39:51 -0700

On Sep 12, 2007, at 12:49 PM, jacob c wrote:

I'm not totally sure I'm following but.. HTTP Reassembly is enabled (checkmarked) under Edit > Prefrences > HTTP if that is what you mean.

Yes, that's what I mean.

I am attaching a screenshot so you can see the display window.

Unfortunately, we need more information than that to debug the problem; if you could give us the full capture file or, at minimum, all the packets in that TCP connection, that'd help (and would probably take less time to download from a mail server than a screenshot, as per Luis's mail).

The replies do show up as "HTTP Continuation" in Ethereal 0.99.0

...which means either that you didn't have HTTP reassembly enabled in 0.99.0 or it wasn't working in 0.99.0 (I forget whether it was in 0.99.0 or not; there have been changes to it since then).

but not in Wireshark 0.99.5 which I am currently using

0.99.5 isn't "the current version of Wireshark"; 0.99.6 is.  Try that.

so perhaps I don't have an option configured correctly. Also, even in Ethereal 0.99.0 it does not decode the WSDL information with or without reassembly enabled.

If it shows up as "HTTP Continuation" in 0.99.0 regardless of whether HTTP reassembly is enabled, it probably means reassembly isn't happening for some reason. Without seeing the packets we can't determine what reason that might have been in 0.99.0 and why the reassembly doesn't finish in 0.99.5.

It just shows up as HTTP data but perhaps Wireshark could decode it once I get it configured correctly. -??

Only if getting it configured correctly means making the reassembly happen correctly. Wireshark doesn't dissect HTTP traffic as anything other than raw data if that traffic isn't part of the first TCP segment of a request or reply and isn't reassembled along with the first segment; that's by design (otherwise, it doesn't know *how* to dissect it - it has to see the Content-Type header, for example).