incubator-wave-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ali Lown <...@lown.me.uk>
Subject Re: Proxy issues
Date Mon, 24 Sep 2012 18:34:19 GMT
If you would like to test it again now/tomorrow?

It took a few hours longer than I expected because I had to stop and
write a patch for Wave (and have dinner, and everything else) to make
it work.

This should have all traffic going over port 443, so if you check in
Wireshark all you should see is some TLS traffic to 71.19.144.245.

Ali

On 24 September 2012 17:18, Ben Hegarty <hegsie@gmail.com> wrote:
> Whenever you get a chance to do that I'll be happy to retest :)
> Thanks again
>
> On Mon, Sep 24, 2012 at 5:14 PM, Ali Lown <ali@lown.me.uk> wrote:
>
>> Yes, packet #46 because I try to make you connect over 9898.
>> (This is because I have the configuration mis-setup, but didn't want
>> to reboot the wave server to fix it).
>>
>> I can move it so that websockets goes over 443, then I will let you
>> try again. (At which time it should work fine).
>>
>> On 24 September 2012 17:09, Ben Hegarty <hegsie@gmail.com> wrote:
>> > https://docs.google.com/open?id=0B5FF_Ld8SzsNMnlmZkZWZWtEQ28
>> >
>> > Looks like you're right there Ali I'm seeing port not allowed in the http
>> > packets
>> > Cheers
>> >
>> > On Mon, Sep 24, 2012 at 5:03 PM, Ali Lown <ali@lown.me.uk> wrote:
>> >
>> >> Yes.
>> >>
>> >> On 24 September 2012 17:01, Ben Hegarty <hegsie@gmail.com> wrote:
>> >> > Sure I can try there too, is it still set with the same dets?
>> >> > Regards
>> >> >
>> >> >
>> >> > On Mon, Sep 24, 2012 at 4:59 PM, Ali Lown <ali@lown.me.uk> wrote:
>> >> >
>> >> >> Extracting the data as raw bytes from the first Websocket response
>> >> >> packet (#95) gives us the following HTML page (attached).
>> >> >>
>> >> >> So, it is _definitely_ an issue with your proxy server not
>> >> >> understanding the Websockets.
>> >> >>
>> >> >> For more information on exactly how they work, a good article would
>> >> >> be: http://lucumr.pocoo.org/2012/9/24/websockets-101/
>> >> >> "The protocol went through many iterations and basically had to
be
>> >> >> changed multiple times because of unforeseen security problems
that
>> >> >> came up with misbehaving proxies." seems to sum-up the problem.
>> >> >>
>> >> >> Ali
>> >> >>
>> >> >> NB: When you tried on my server (https://wave.eezysys.co.uk), I
am
>> >> >> less certain as to why it failed there given all the traffic is
>> >> >> encrypted. (Unless your company proxy is terminating my SSL
>> >> >> connection, performing DPI on the now-decrypted data, and then
>> >> >> re-encrypting it before presenting it to you)
>> >> >> Could you do a wireshark capture for that server as well?
>> >> >> Actually, it might be because my server still tries to use a
>> >> >> non-standard port for the websockets, and it is quite likely you
have
>> >> >> most outgoing ports blocked.
>> >> >>
>> >> >> On 24 September 2012 16:42, Ben Hegarty <hegsie@gmail.com>
wrote:
>> >> >> > Hey Ali,
>> >> >> > Basically I get 'A turbulance' after logging in and never
go online
>> >> and
>> >> >> no
>> >> >> > wave data is saved down, you just see 'Unsaved all the time'..
>> >> >> > I've uploaded the wireshark trace to the following location
:)
>> >> >> >
>> >> >> > https://docs.google.com/open?id=0B5FF_Ld8SzsNMm5oOGJXajlOV00
>> >> >> >
>> >> >> > HTH
>> >> >> >
>> >> >> > On Mon, Sep 24, 2012 at 4:33 PM, Ali Lown <ali@lown.me.uk>
wrote:
>> >> >> >
>> >> >> >> > basically it looks like the problem comes from the
fact that the
>> >> >> request
>> >> >> >> > uses a relative path instead of a full path for the
request
>> >> >> >>
>> >> >> >> AFAIK it is the browser's responsibility to convert relative
paths
>> >> for
>> >> >> >> communicating to the server.
>> >> >> >>
>> >> >> >> > I've been investigating some issues that I've been
having with
>> >> using
>> >> >> wave
>> >> >> >> > behind a proxy server
>> >> >> >>
>> >> >> >> What are the actual problems you are having? Unable to
get
>> 'Online'
>> >> >> >> because the websocket doesn't get established?
>> >> >> >>
>> >> >> >> > and found this thread about proxy issues which are
>> >> >> >> > identical to the issue I'm seeing...
>> >> >> >> > https://bugs.eclipse.org/bugs/show_bug.cgi?id=107696
>> >> >> >>
>> >> >> >> This seems to be about eclipse's internal proxy settings
handling
>> >> >> >> being incorrect when it parses the relative URL. Assuming
your
>> >> browser
>> >> >> >> is setup correctly the problem shouldn't lie with that.
>> >> >> >>
>> >> >> >> > This GET here looks to be the source of the problem,
I'm no
>> network
>> >> >> guy
>> >> >> >> so
>> >> >> >> > it would be nice to have this confirmed by someone
else, but do
>> you
>> >> >> think
>> >> >> >> > this would be a difficult fix? I have my own env
for testing
>> this
>> >> so I
>> >> >> >> > could easily apply a patch and retest just need some
help with
>> the
>> >> >> >> > patch.. *(fingers
>> >> >> >> > crossed this is the issue)*
>> >> >> >>
>> >> >> >> The normal problem with websockets failing that I have
seen, is
>> when
>> >> >> >> the proxy server doesn't understand (so drops) the UPGRADE
>> request.
>> >> >> >>
>> >> >> >> Alternatively, it may well be the relative path, in which
case the
>> >> >> >> problem lies with your proxy server being incorrectly
configured
>> to
>> >> >> >> handle them (since relative paths are most definitely
part of
>> >> >> >> RFC2616).
>> >> >> >>
>> >> >> >> What is the proxy server in question?
>> >> >> >>
>> >> >> >> It may be worth using wireshark to watch the actual request
across
>> >> the
>> >> >> >> wire.
>> >> >> >>
>> >> >> >> Ali
>> >> >> >>
>> >> >> >
>> >> >> >
>> >> >> >
>> >> >> > --
>> >> >> > Mobile Phone: +447767-322-122
>> >> >> > Work Phone: +4420 79485612
>> >> >>
>> >> >
>> >> >
>> >> >
>> >> > --
>> >> > Mobile Phone: +447767-322-122
>> >> > Work Phone: +4420 79485612
>> >>
>> >
>> >
>> >
>> > --
>> > Mobile Phone: +447767-322-122
>> > Work Phone: +4420 79485612
>>
>
>
>
> --
> Mobile Phone: +447767-322-122
> Work Phone: +4420 79485612

Mime
View raw message