Ian Sorry missed your email. Yes H245Routed would help. This allows the H.225 and H.245 to be routed through the gatekeeper so support H.245 tunneling in H.225. 1 TCP connection. Under the other setting the H.225 is gatekeeper routed but the media control goes directly bypassing the gatekeeper. So 2 separate TCP connections. H.245 gets a lot more activity during a call. (FastPictureUpdate,FlowControl etc) TCP connections in routers generally timeout in between 20-45 sec horrible but true. The magic number we use for H.460.18/.19 is always 19sec. Forcing a TCP keepAlive in the OS definitely helps. Simon From: Ian Blenke [mailto:ian@xxxxxxxxxx]
Sent: 30 November 2011 07:47
To: GNU Gatekeeper Users
Subject: Re: Question about disengageRequest Simon, might this help him? H245Routed=1
I hadn't even thought of the TCP session timing out that quickly. UDP, definitely, but TCP timing out that fast just doesn't seem right to me (though it wouldn't be entirely suprising). Alternatively, if that were the case, would this help? sysctl -w net.ipv4.tcp_keepalive_time=15 By default, that's 7200 (2 hours), not 15 seconds. - Ian On Tue, Nov 29, 2011 at 4:34 PM, Simon Horne <s.horne@xxxxxxxxxxxxxx> wrote: Mikael
My guess is that the H.225 TCP connection is being dropped forcing the DRQ
from GnuGk (shutdown the call). Hence the forceDrop reason
Generally firewalls will drop TCP connections after about 20 sec of
inactivity. The H.225 channel gets very little activity if not H.245
tunneled. Make sure there is nothing between the devices that could do that.
The only other reason is the Mirial side is deciding to drop the TCP
connection due to a timeout or something. You will need to check the config
and logs as well. You might need to consult mirial.
Either case the error is coming from outside GnuGk control.
Simon
-----Original Message-----
From: Mikael Grehn [mailto:mikael.grehn@xxxxxxxx]
Sent: 30 November 2011 07:17
To: GNU Gatekeeper Users
Subject: Re: Question about disengageRequest
Thank you Jan!
My main skill is unfortunately SIP but on that side I can see the SIP
Invite, SIP OK and later SIP Ack which should mean that H323 has connected
the call?
I have tried to go through the Gnugk traces but have difficulties detecting
the source of the error.
I see:
maintainConnection = false
supportsAltGK = <<null>>
in the registration that looks bad.
In the sudden disengageRequest (from primary gnugk directly to Mirial) I see
a forceDrop as disengageReason. Without getting anything from the H323
client?
-------------
2011/11/29 16:12:56.201 3 RasSrv.cxx(235) RAS Send to
207.183.235.149:1719
disengageRequest {
requestSeqNum = 2
endpointIdentifier = 9 characters {
0036 0039 0035 0037 005f 0065 006e 0064 6957_end
0070 p
}
conferenceID = 16 octets {
02 35 81 4e 9a 4b 71 12 36 b9 e2 73 30 a9 2e cf .5.N.Kq.6..s0...
}
callReferenceValue = 51151
disengageReason = forcedDrop <<null>>
callIdentifier = {
guid = 16 octets {
02 35 81 4e 9a 4b 71 12 36 b8 e2 73 30 a9 2e cf .5.N.Kq.6..s0...
}
}
gatekeeperIdentifier = 6 characters {
0047 0047 006e 0075 0047 006b GGnuGk
}
answeredCall = false
}
-------------
If it is OK with you I'd like to send you the gnugk traces directly (not
exposing real IP's), both the primary and alternative one together with my
gatekeeper configuration?
Only if that is OK of course! :)
Best Regards,
Mikael
-----Ursprungligt meddelande-----
Från: Jan Willamowius [mailto:jan@xxxxxxxxxxxxxx]
Skickat: den 29 november 2011 20:55
Till: openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Ämne: Re: Question about disengageRequest
Again, audio and video is no indication that the call has connected (and
would be past most of the message timeouts). You have to check for the
Connect message and check if the H.245 connection negotiated properly.
Regards,
Jan
Mikael Grehn wrote:
> Hi Jan and thanks for the rapid response.
>
> The calls are perfectly connected and they run with audio and video for
about 25 seconds (perhaps 20 seconds after fully being connected). Then the
primary gatekeeper is sending this disengageRequest to Mirial and Mirial
disconnects both sides. Primary gatekeeper has been "quiet" for about 20
seconds before this so something is absolutely happening!
>
> If I make a call to the other customer (directly to primary Gk) the
> call is not disconnected at all and are working perfectly. :)
>
> Any ideas what can trigger this?
>
> Thanks for you kind help!
>
> Best Regards,
> Mikael
>
>
> -----Ursprungligt meddelande-----
> Från: Jan Willamowius [mailto:jan@xxxxxxxxxxxxxx]
> Skickat: den 29 november 2011 20:21
> Till: openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
> Ämne: Re: Question about disengageRequest
>
> Hi Mikael,
>
> are your calls being terminated before or after they connect ?
> (Hearing voice may just mean that the call was using FastStart.)
>
> If GnuGk actively terminates a call it usually detected ran into a timeout
(it tries to detect stale calls). When you have multiple gatekeepers
involved, make sure each gatekeeper that handles the call sees all messages
for that call.
>
> Regards,
> Jan
>
>
> Mikael Grehn wrote:
> > Hi!
> >
> > I have a question about why a primary gatekeeper(out of 2
> > gatekeepers) is disconnecting calls after 25 seconds that first
> > reaches the alternative gatekeeper on another machine. The H323
> > endpoint that is receiving the call is a Mirial gateway. :)
> >
> > We have two customers that use incoming H323 and SIP for their service.
> >
> > For the H323 we are using a H323 Gnugk for each customer (on two
different IP's) that I have tried to setup as primary and alternative
gatekeeper.
> > To translate from H323 to SIP (which is the protocol mainly
> > supported by our service) we have a Mirial H323 2 SIP in between.
> > The H323 side of Mirial is registered in the primary gatekeeper at
> > customer 1. We cant register it in the alternative as well. :)
> >
> >
> > Customer 1: GnuGk(primary) IP1 <--registered Mirial IP3
> > |reg
> > Customer 2: GnuGk(alternative) IP2
> >
> >
> > Calls from H323 clients to the first (primary) customer is working
perfectly.
> > Calls from H323 to the second customer (with the alternative gatekeeper
gets the call to work for about 25 seconds. After this the primary
gatekeepers decides to send disengageRequest directly to Mirial gateway,
that is later disconnecting the calls.
> >
> > This is the log from the primary gatekeeper (207.183.111.122) when it is
disconnecting a working call:
> >
> > (. . . about 20 seconds with no logs since the call was setup OK . .
> > .) disengageRequest {
> > requestSeqNum = 104
> > endpointIdentifier = 9 characters {
> > 0031 0034 0030 0038 005f 0065 006e 0064 1408_end
> > 0070 p
> > }
> > conferenceID = 16 octets {
> > 02 35 81 01 a4 43 d0 10 36 b3 e2 73 30 a9 2e cf
.5...C..6..s0...
> > }
> > callReferenceValue = 51145
> > disengageReason = forcedDrop <<null>>
> > callIdentifier = {
> > guid = 16 octets {
> > 02 35 81 01 a4 43 d0 10 36 b2 e2 73 30 a9 2e cf
.5...C..6..s0...
> > }
> > }
> > gatekeeperIdentifier = 6 characters {
> > 0047 0047 006e 0075 0047 006b GGnuGk
> > }
> > answeredCall = false
> > }
> > 2011/11/29 12:55:55.689 2 RasTbl.cxx(2642) Gk
Disconnect Call No. 26
> > 2011/11/29 12:55:55.689 2 RasTbl.cxx(4076) CDR ignore
not connected call
> > 2011/11/29 12:55:55.689 2 gkacct.cxx(955) GKACCT
Successfully logged event 2 for call no. 26
> > 2011/11/29 12:55:55.690 2 RasSrv.cxx(173) RAS Read
from 207.183.111.224:1719
> > 2011/11/29 12:55:55.690 3 RasSrv.cxx(223) RAS
> > disengageConfirm {
> > requestSeqNum = 104
> > usageInformation = {
> > nonStandardUsageFields = 0 entries {
> > }
> > alertingTime = 1322592922
> > connectTime = 1322592926
> > }
> > }
> > 2011/11/29 12:55:55.690 1 RasSrv.cxx(355) RAS DCF
Received from 207.183.111.224:1719
> > 2011/11/29 12:55:56.689 3 RasTbl.cxx(2130) Gk Delete
Call No. 26
> > 2011/11/29 12:55:57.278 2 RasSrv.cxx(173) RAS Read
from 207.183.111.224:1719
> > 2011/11/29 12:55:57.278 3 RasSrv.cxx(223) RAS
> > disengageRequest {
> > requestSeqNum = 60528
> > endpointIdentifier = 9 characters {
> > 0031 0034 0030 0038 005f 0065 006e 0064 1408_end
> > 0070 p
> > }
> > conferenceID = 16 octets {
> > 02 35 81 01 a4 43 d0 10 36 b3 e2 73 30 a9 2e cf
.5...C..6..s0...
> > }
> > callReferenceValue = 18377
> > disengageReason = normalDrop <<null>>
> > callIdentifier = {
> > guid = 16 octets {
> > 02 35 81 01 a4 43 d0 10 36 b2 e2 73 30 a9 2e cf
.5...C..6..s0...
> > }
> > }
> > gatekeeperIdentifier = 6 characters {
> > 0047 0047 006e 0075 0047 006b GGnuGk
> > }
> > answeredCall = true
> > usageInformation = {
> > nonStandardUsageFields = 0 entries {
> > }
> > alertingTime = 1322592922
> > connectTime = 1322592926
> > endTime = 1322592952
> > }
> > terminationCause = releaseCompleteReason gatekeeperResources
<<null>>
> > }
> > 2011/11/29 12:55:57.279 1 RasSrv.cxx(355) RAS DRQ
Received from 207.183.111.224:1719
> > 2011/11/29 12:55:57.279 2 RasSrv.cxx(395)
DCF|207.183.111.224|1408_endp|18377|normalDrop|02-35-81-01-a4-43-d0-10-36-b2
-e2-73-30-a9-2e-cf;
> > 2011/11/29 12:55:57.279 3 RasSrv.cxx(235) RAS Send to
207.183.111.224:1719
> > disengageConfirm {
> > requestSeqNum = 60528
> > }
> >
> > Any ideas why it does this and how I can configure it away?
> > I am adding the gnugk.ini file from primary gatekeeper. It is very
> > similar on the alternative one. :)
> >
> > Primary Gk: 207.183.111.222
> > Alternativ Gk: 207.183.111.223
> > Mirial h323 2 SIP gw: 207.183.111.224
> >
> > [Gatekeeper::Main]
> > FortyTwo=42
> > Name=GGnuGk
> > TimeToLive=600
> > StatusTraceLevel=2
> > AlternateGKs=207.183.111.223:1719:false:120:CGnuGk
> > SendTo=207.183.111.223:1719
> > SkipForwards=207.183.111.223
> > CompareAliasCase=0
> > CompareAliasType=0
> > ExternalIP=207.183.111.222
> > [GkStatus::Auth]
> > rule=allow
> > 127.0.0.1=allow
> > default=allow
> > Shutdown=allow
> >
> > [RoutedMode]
> > GKRouted=1
> > H245Routed=0
> > CallSignalPort=1720
> > AcceptUnregisteredCalls=1
> > DropCallsByReleaseComplete=1
> > SupportNATedEndpoints=1
> >
> > [Proxy]
> > Enable=0
> > ProxyForNAT=1
> > ProxyForSameNAT=0
> >
> > [RoutingPolicy]
> > default=explicit,internal,parent,neighbor,catchall,vqueue
> >
> > [RasSrv::RewriteAlias]
> > customerdomain.com=0099999999
> >
> > [Routing::CatchAll]
> > CatchAllAlias=PSEGateway
> >
> > [EP::PSEGateway]
> > Capacity=30
> > AddNumbers=0099999999
> > AdditionalDestinationAlias=0099999999
> >
> > I have tried to change the real IP's into fictional ones, might have
missed some!
> > Any help is very appreciated! :)
> >
> > Best Regards,
> > Mikael
--
Jan Willamowius, jan@xxxxxxxxxxxxxx, http://www.gnugk.org/
----------------------------------------------------------------------------
--
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security threats,
fraudulent activity, and more. Splunk takes this data and makes sense of it.
IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________________
Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive:
http://sourceforge.net/mailarchive/forum.php?forum_name=openh323gk-users
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
----------------------------------------------------------------------------
--
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security threats,
fraudulent activity, and more. Splunk takes this data and makes sense of it.
IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________________
Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive:
http://sourceforge.net/mailarchive/forum.php?forum_name=openh323gk-users
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure
contains a definitive record of customers, application performance,
security threats, fraudulent activity, and more. Splunk takes this
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________________
Posting: mailto:Openh323gk-users@xxxxxxxxxxxxxxxxxxxxx
Archive: http://sourceforge.net/mailarchive/forum.php?forum_name=openh323gk-users
Unsubscribe: http://lists.sourceforge.net/lists/listinfo/openh323gk-users
Homepage: http://www.gnugk.org/
--
- Ian Blenke <ian@xxxxxxxxxx> http://ian.blenke.com |
