Well, I've got a sample split "rcon status" response packet that does NOT
have the same split behavior as usual (or what I expected).  I have the
entire packet contents of both parts.  I'd like to post it here, but I know
some people would not like having their IPs and WON IDs made public, so is
there someone I can send it to directly?

The first part was 1380 bytes and the second part was 437.  I also noticed
that the packet was split "cleanly" at the end of a line.  I suspect there
may be some code that checks to see if it can split at a line boundary and
just send it as two "normal" packets.  Could this be the case?

Each part starts with "FF FF FF FF 6C".  Also, I know the MTU size of my
router and the hlds server at the other send is set to 1500, but I don't
know about in-between.  If I send an ICMP ping with data, I find I can send
a max of 1440 bytes before things get split, which seems about right to me.

Let me know where I can send the contents of those packets.  It was pretty
reproducible tonight - about 2 out of 3 "rcon status" commands resulted in
this behavior.

> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Kris
> Sent: Monday, December 30, 2002 12:06 PM
> To: [EMAIL PROTECTED]
> Subject: Re: [hlds_apps] Bug report, question on packet splitting &
> secure
>
>
> At 03:33 27/12/2002, you wrote:
> >hmm, can you say how to replicate this? (i.e start a server with
> "x" in the
> >command line, then type "y" 10 times, blah...)
>
> Easy - Start up a server and get 24 clients to join, then do a
> remote 'rcon
> status' :)
>
> >I have a similar program which uses rcon extensively and have never seen
> >this behaviour... I would like to replicate it :)
>
> You could probably replicate this by lowering the MTU (Maximum
> Transmission
> Unit) size from within your OS. (note, this may drastically affect your
> network throughput if set too small).
>
> If we could get some sort of packet ID in the response, or if half-life
> could be clever and split up packets based on the host OS's MTU that'd be
> great...
>
> Kris.
>
> >----- Original Message -----
> >From: "Terry" <[EMAIL PROTECTED]>
> >To: <[EMAIL PROTECTED]>
> >Sent: Wednesday, December 25, 2002 1:58 AM
> >Subject: RE: [hlds_apps] Bug report, question on packet
> splitting & secure
> >
> >
> > > I'd like to resurrect this again.  :-)
> > >
> > > I've changed my code to work with this so that split packets get
> >reassembled
> > > and properly ordered.  This works *most* of the time, but I find that
> > > sometimes responses to commands like "rcon status" are split
> but do not
> >use
> > > this convention.
> > >
> > > For example, if there are 24 players on my server, usually
> the response to
> > > an "rcon status" is split.  Sometimes, it uses the convention
> below and I
> > > can reassemble and reorder it.  But sometimes it just looks like two
> > > seperate "normal" packets, i.e. each packet starts with "ff
> ff ff ff 6C
> > > <data>".
> > >
> > > I'm guessing that it only does this if the break happens to
> fall at the
> >end
> > > of a line.  (??)  If so, is this a bug or intended behavior?
> I personally
> > > would rather it *always* split the same way because my
> program is trying
> >to
> > > parse what comes back but I have no way of knowing if/when
> all the data is
> > > there.
> > >
> > > Terry
> > >
> > > > -----Original Message-----
> > > > From: [EMAIL PROTECTED]
> > > > [mailto:[EMAIL PROTECTED]]On Behalf Of Alfred
> > > > Sent: Thursday, November 21, 2002 1:14 PM
> > > > To: [EMAIL PROTECTED]
> > > > Subject: Re: [hlds_apps] Bug report, question on packet splitting &
> > > > secure
> > > >
> > > >
> > > > [EMAIL PROTECTED] wrote:
> > > > <snip>
> > > >
> > > > > >
> > > > > >first- splitted udp packets if the response of a query
> gets to long
> > > > > >(discussed here one than one time.)
> > > > > >
> > > > > >I found this analysis - describing the rules response -  anywhere
> > > > > (qstat I
> > > > > >think):
> > > > > >
> > > > > >splitted to two packets:
> > > > > >FF FF FF FF 6C [data]
> > > > > >FE FF FF FF 07 02 00 00 02 FF FF FF FF 6C [data]
> > > > > >FE FF FF FF 07 02 00 00 12 [data]
> > > > > >
> > > > > >splitted to three packets:
> > > > > >FE FF FF FF 07 02 00 00 03 FF FF FF FF 6C [data]
> > > > > >FE FF FF FF 07 02 00 00 13 [data]
> > > > > >FE FF FF FF 07 02 00 00 23 [data]
> > > > > >
> > > > > >07 02 00 00 variates and is some kind of challenge numer - an
> > > > id telling
> > > > > >which packets schould be one.
> > > > > >the next byte is splittet into two 4 bit nibles - the upper
> > > > one describes
> > > > > >the packet number, the lower one how many packets will be sent.
> > > > > >
> > > > > >is this correct???
> > > > > >
> > > > > >
> > > > > >
> > > >
> > > > <snip>
> > > >
> > > > No, its not quiet right. Split packets are denoted by an
> initial FF FF
> > > > FF FE tag (rather than FF FF FF FF) (that is, decimal -2 rather than
> > > > decimal -1). The next byte in a split packet contains 2 4
> bit nibbles
> > > > (mmm, nibbles), with the first 4 bits being the  packet
> number and the
> > > > last 4 being the total number of packets.
> > > > If the packet is the first in the sequence it will then
> have a FF FF FF
> > > > FF tag on the first packet and then data (i.e the long
> packet was simply
> > > > "packed" inside of multiple shorter packets).
> > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > hlds_apps mailing list
> > > > [EMAIL PROTECTED]
> > > > http://list.valvesoftware.com/mailman/listinfo/hlds_apps
> > > >
> > >
> > >
> > >
> > > _______________________________________________
> > > hlds_apps mailing list
> > > [EMAIL PROTECTED]
> > > http://list.valvesoftware.com/mailman/listinfo/hlds_apps
> > >
> >
> >_______________________________________________
> >hlds_apps mailing list
> >[EMAIL PROTECTED]
> >http://list.valvesoftware.com/mailman/listinfo/hlds_apps
>
> _______________________________________________
> hlds_apps mailing list
> [EMAIL PROTECTED]
> http://list.valvesoftware.com/mailman/listinfo/hlds_apps



_______________________________________________
hlds_apps mailing list
[EMAIL PROTECTED]
http://list.valvesoftware.com/mailman/listinfo/hlds_apps

Reply via email to