The ITU specs do not define behaviour when processing LRQ without hopCount field.
Currently, GnuGk behaves like it would be LRQ with hopCount = 1. It check its internal
registration table whether the destination is registered at this gatekeeper, but does not forward
LRQ further.
But maybe it will be a better solution to include hopCount field when it is not present
in the LRQ received, before forwarding it to neighbors. This will prevent endless LRQ loops.
I would modify the patch further to set hopCount = ForwardHopCount, when ForwardLRQ=always
or ForwardLRQ=depends and hopCount field is missing.
----- Original Message -----
From: "Thomas" <thomasj@xxxxxxxxx>
Sent: Sunday, March 21, 2004 8:51 AM
> Unfortunately I don't have access to ITU specs so I can't check what
> is the suggested policy for hopCount.
> Is hopCount a used field in LRQs? In my example GK1 was also a GnuGk
> gatekeeper. What do I have to set up for it to include hopCount field
> in LRQ requests? [ForwardHopCount=2] Why it is not sent always?
> Do somebody know that other gatekeepers (like Cisco) fill/set this
> field in LRQs or not?
> If it is not strictly defined [in standard] we could run into some
> pain situation, when the sending gatekeeper does not have hopCount
> field and it's not possible set it up.
> (I found hopcount setting in Cisco gatekeeper:
> "Examples
>
> The following example shows a maximum of 10 hops through which an address resolution can be forwarded.
>
> Router(config)# call-router h323-annexg be20
> Router(config-annexg)# hopcount 10"
> Should it be this the right setting?)
>
> Thanks in advance,
> Thomas
>
>
> Saturday, March 20, 2004, 4:54:33 PM, Zygmuntowicz wrote:
> ZM> I think if the hopCount field is missing, the LRQ should not be forwarded,
> ZM> so the original code is working well in this case. Any other opinions?
>
> ZM> ----- Original Message -----
> ZM> From: "Thomas" <thomasj@xxxxxxxxx>
> ZM> Sent: Saturday, March 20, 2004 1:21 AM
>
>
> >> The bug description was sent to the users list, however after looking
> >> for the source codes I found that this is not a configuration issue
> >> but a bug in code (most probably).
> >>
> >> Short description:
> >> The ForwardHopCount option was not used in routing decision.
> >>
> >> Longer description:
> >>
> >> bool NeighborPolicy::OnRequest(LocationRequest & lrq_obj)
> >> {
> >> RasMsg *ras = lrq_obj.GetWrapper();
> >> List::iterator iter = find_if(m_neighbors.begin(),
> >> m_neighbors.end(), bind2nd(mem_fun(&Neighbor::IsAcceptable), ras));
> >> Neighbor *requester = (iter != m_neighbors.end()) ? *iter : 0;
> >> int hopCount = 0;
> >> if (requester)
> >> if (requester->ForwardLRQ() < 0)
> >> return false;
> >> else if (requester->ForwardLRQ() > 0)
> >> hopCount = 1;
> >>
> >> H225_LocationRequest & lrq = (*ras)->m_recvRAS;
> >> if
> >> (lrq.HasOptionalField(H225_LocationRequest::e_hopCount)) {
> >> hopCount = lrq.m_hopCount - 1;
> >> if (hopCount)
> >> lrq.m_hopCount = hopCount;
> >> }
> >> if (!hopCount)
> >> return false;
> >>
> >> If I set ForwardLRQ=always, I had a good workaround. The default
> >> ForwardLRQ's values is depends, which is not handled in previouse
> >> code. So hopCount=0.
> >> The LRQ request had no hopCount field, so the hopCount variable
> >> remained 0, thus the OnRequest function returned with false.
> >>
> >> Here is a simple bugfix (I'm not sure if it is 100% correct) agains
> >> anon. CVS (2_2):
> >> cvs diff -u Neighbor.cxx
> >> Index: Neighbor.cxx
> >> ===================================================================
> >> RCS file: /cvsroot/openh323gk/openh323gk/Neighbor.cxx,v
> >> retrieving revision 1.19
> >> diff -u -r1.19 Neighbor.cxx
> >> --- Neighbor.cxx 9 Mar 2004 11:36:42 -0000 1.19
> >> +++ Neighbor.cxx 19 Mar 2004 23:57:54 -0000
> >> @@ -853,6 +853,11 @@
> >> return false;
> >> else if (requester->ForwardLRQ() > 0)
> >> hopCount = 1;
> >> + else {
> >> + hopCount =
> >> requester->GetDefaultHopCount() - 1;
> >> + if (hopCount < 0)
> >> + hopCount = 0;
> >> + }
> >>
> >> H225_LocationRequest & lrq = (*ras)->m_recvRAS;
> >> if
> >> (lrq.HasOptionalField(H225_LocationRequest::e_hopCount)) {
> >>
> >> Kind regards,
> >> Thomas
> >>
> >> Friday, March 19, 2004, 8:57:25 PM, Thomas wrote:
> >> T> Hello,
> >>
> >> T> I have 3 GKs interconnected theirselves with neighboring.
> >> T> GK1-GK2-GK3
> >> T> I would like to forward LRQs on GK2 coming from GK1 to GK3. I set up
> >> T> routing but somehow it doesn't work.
> >> T> 2004/03/19 20:17:27.289 5 job.cxx(186) JOB
> >> T> Starting Job LRQ at Worker thread 9223
> >> T> 2004/03/19 20:17:27.289 1 RasSrv.cxx(306) RAS LRQ Received
> >> T> 2004/03/19 20:17:27.289 2 Toolkit.cxx(291)
> >> T> RewritePString: 111111 to 1212#1111
> >> T> 2004/03/19 20:17:27.289 4 gkauth.h(113) GkAuth default check ok
> >> T> 2004/03/19 20:17:27.289 5 Routing.h(117) ROUTING
> >> T> Checking policy Internal for the request LRQ 59
> >> T> 2004/03/19 20:17:27.289 5 Routing.h(117) ROUTING
> >> T> Checking policy Neighbor for the request LRQ 59
> >> T> 2004/03/19 20:17:27.289 2 RasSrv.cxx(351)
> >> T> LRJ|<gk1ip>|1212#111:dialedDigits|<gk1id>|requestDenied
> >>
> >> T> Call from an endpoint registered under GK2 goes normally to GK3
> >> T> through LRQ, so I don't think it's the problem. Also, I can call from
> >> T> an endpoint registered to GK1 an endpoint registered under GK2.
> >>
> >> T> Configuration on GK2:
> >> T> [RasSrv::LRQFeatures]
> >> T> NeighborTimeout=2
> >> T> IncludeDestinationInfoInLCF=1
> >> T> CiscoGKCompatible=1
> >> T> ForwardHopCount=2
> >> T> AcceptForwardedLRQ=1
> >> T> ForwardResponse=1
> >>
> >> T> [Neighbor::GK3]
> >> T> Host=<gk3ip>
> >> T> Dynamic=0
> >> T> SendPrefixes=1212#
> >> T> AcceptForwardedLRQ=1
> >>
> >> T> [Neighbor::GK1]
> >> T> Host=<gk1ip>
> >> T> Dynamic=0
> >> T> AcceptPrefixes=1212#
> >> T> AcceptForwardedLRQ=1
> >> T> ForwardHopCount=2
> >>
> >> T> [RoutingPolicy]
> >> T> Default=internal,neighbor,parent
> >>
> >> T> Thanks in advance,
> >> T> Thomas
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
List: Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549
Homepage: http://www.gnugk.org/
