Clarification on Gateway Commands

[Dass Prakash R]

Hi Experts,

I have a few clarifications on Gateways. Kindly provide your valuable points

Let me consider I have two dial-peers

say for example:

dial-peer voice 1 voip
destination-pattern .T
Session-Target IPv4: 1.1.1.1
Preference 1
Session Protocol sipv2

dial-peer voice 2 voip
destination-pattern .T
session-Target IPV4: 1.1.1.2
Preference 2
Session Protocol sipv2

The Gateway chooses Preference 1 if 1.1.1.1 is alive or it moves to preference 2.

My Question here is how does the Gateway checks the sesison Target is alive or not in SIP platform. What kind of messages is being interchanged.

If I consider the scenario, that 1.1.1.1 is down and calls are flowing thro Preference 2 and If the IP 1.1.1.1 comes alive how much time the gateway will take to switch back to preference 1. Can we also configure the timings?

What is the Use of the following commands in the Gateway

bind control source-interface Gigabitethernet 0/0
bind media source-interface Gigabitethernet 0/0

If I do not bind the interface if I have only one interface configured what will be the effect on the Gateway. Will I face difficulties during any failover scenarios?

What is the Use of the following commands in the Gateway
min-se 360 session-expires 360

Thanks in advance

Regards,

Dass

[Jonathan Schulenberg · Accepted solution]

IOS will send a SIP INVITE message to start the call and waits for the 100 TRYING response. If it doesn't receive one it will resend the INVITE message until the maximum retries have been reached. When it reaches the maximum and if the retries have been set to four or fewer (non-default setting!) it will proceed to the next dial-peer.

sip-ua 
 retry invite 2

It does not maintain keepalives with the SIP peer by default. This is possible using SIP OPTIONS PING which sends SIP messages and expects a Layer 7 response (not ICMP). You can add this to the dial-peer; however, it's unidirectional. If you want the far-end to also maintain a state you would need to enable OPTIONS PING from that side as well. In CUCM this was added in CUCM 8.5(1) and is set under the SIP Profile.

dial-peer voice 1 voip
 voice-class sip options-keepalive

As a side-node IOS will not stop routing the call if it receives a 400 responce from the peer (e.g. 486 Unassigned Number). It would proceed to the second dial-peer and try again. Since the second dial-peer leads to the same cluster you can add the command huntstop to both dial-peers as well. When combined with the options-keepalive command above this will prevent IOS from hunting beyond the first matching dial-peer when a 400 error is returned. When the keepalives busyout the first dial-peer IOS will then choose the second one.

What is the Use of the following commands in the Gateway
bind control source-interface Gigabitethernet 0/0
bind media source-interface Gigabitethernet 0/0
If I do not bind the interface if I have only one interface configured what will be the effect on the Gateway. Will I face difficulties during any failover scenarios?

This determines the source IP address used for the SIP and RTP traffic. Without the bind command the router will set the source IP to whatever interface the routing table chooses for the packet to egresses from the router. It typically (but not always in the case of CUBE) is set to a loopback interface to ensure that the VoIP functionality functions even when an individual interface fails. If the interface the process is bound to goes down the VoIP functionality stops as well.

What is the Use of the following commands in the Gateway
min-se 360 session-expires 360

You can always use the Command Reference Guide to lookup a command. In this case it's a reference to SIP mid-session keepalives (which otherwise do not exist by default as they do with H.225). The keepalive mechanism is explained in RFC4028. This allows the router or it's peer to tear down the call should the other fail to respond to a mid-session re-INVITE message.

Please remember to rate helpful responses and identify helpful or correct answers.

[Dass Prakash R]

Hi Jonathan,

Thank you a lot for the reply.

To drill down further,

1. Can I bind a virtual interface i,e for example bind control source-interface Gigabitethernet 0/0.448

2. I also have an issue. I have two Gateways separated over WAN. They are redundant to eachother. In our case let me consider gateway 1 is in City A(Preference 1) and gateway 2 is in City B (Preference 2).

The call flow is, Calls terminate in PSTN Gateway from City C and the call is forwarded to City A as primary and City B as Secondary over WAN

If the WAN link between City C to City A goes down auto failover to City B is happening. But if the WAN link comes up again the Failover back to City A does not happen instantly. Even though the call comes and hits the Gateway the RTP path is not getting established between Cusotmer and the user.

When I checked the CUBE configuration of City C there I do not have the bind command on SIP. Does this failover issue happen because of the Binding command.

To add upon, I have configured rel1xx disable in the dial-peer

Please advice

Thanks in advance

Dass

[Jonathan Schulenberg]

1. Yes.

2. What do you mean "does not happen instantly"? Any connected calls would remain C-A. New calls would be C-B if there is no keepalive in place. If there is you would need the keepalives to notice that it's now working again and to put the dial-peer back in service. show dial-peer voice brief will show the status of the dial-peers.

Also, rel1xx should be left enabled unless you know of a specific reason to disable it.

Please remember to rate helpful responses and identify helpful or correct answers.

[Dass Prakash R]

Hi Jonathan,

Primary path is C-A and if there is no connectivity from C-A then new calls would be C-B. This happens instantly. Also, when the WAN link is up again between C-A, the dial-peer switches back to C-A instantly. I can see the hit in the gateways. However, there is no RTP getting established between the Gateways. In the Phone I could see the call is connected but I could not hear the voice from the other end.

Can you suggest whether the above stated problem is due to the absence of binding in the CUBE Gateway in City C

Thanks in advance.

Regards,

Dass