MBG not failing over if WAN link down

[welshy67]

MBG: 8.1.25.0
vMCD: 7.0 SP1

We have 50 SIP trunks with a SIP provider in Active/Standby mode. We have a Cluster of two MBG's in Server-Gateway mode. Both MBG's have two NIC's (WAN & LAN) on network edge. We have a single vMCD for SIP use at present. ARS digits dialed configured as a List to use MBG1 as 1st choice and MBG2 as 2nd choice. Calls go through MBG1 to SIP cloud 1 & MBG2 to SIP cloud 2 successfully when tested, that is changing ARS digits dialed to a route rather than a list and testing both SIP circuits.

We are still testing but due to port our numbers over from ISDN soon. If I shut down the MBG1 LAN port on our Cisco switch, then after a few seconds calls correctly get routed to the 2nd choice route of MBG2. On vMCD a SIP LINK ACTIVE ALL correctly shows SIP1 as Out of Service. If I re-enable the LAN port, calls re-route back to MBG1.

The problem we have is when I shut down the MBG WAN port to simulate our main SIP circuit failing. This time it takes about 5 minutes for calls to be routed to MBG2, but that's not all, it only works once. If I re-enable the WAN port, calls go back through MBG1 but if I do the whole process again by shutting down the WAN port, calls never go to MBG2 until I reboot MBG1. And a SIP LINK ACTIVE ALL never shown MBG1 out of service.

It seems that MBG and vMCD are not communicating properly when the WAN link fails. I have messed about with the Timers in the SIP Peer Profile and the SIP Keepalives in MBG but nothing works. Our Mitel support Engineers are also scratching their heads, they even passed it onto Mitel who suggested a Timer change but this made it worse as now calls never get routed to MBG2.

vMCD, SIP Peer Profile Timers:

Keep-Alive (OPTIONS) Period  60
Registration Period  3600
Registration Period Refresh (%)  50
Registration Maximum Timeout  60
Session Timer  90
Session Timer: Local as Refresher  No
Subscription Period  3600
Subscription Period Minimum  300
Subscription Period Refresh (%)  80
Invite Ringing Response Timer  8  (this is the change suggested by Mitel)

MBG SIP Options

SIP support:  Enabled
PRACK support:  Disabled
Local streaming: False
Send options keepalives: Always
Options interval: 10
Gap register: True
Set-side registration expiry time: 240
ICP-side registration expiry time: 900
Challenge methods: invite subscribe
Allowed URI names:


Any ideas?

Jon.

[bluewhite4]

I guess my first question/suggestion would be to get a Wireshark trace of the SIP traffic when MBG1 looses its WAN connection for the second time. Maybe from that everyone would be able to determine why the 3300 isn't failing over to MBG2.

[welshy67]

I'll see what I can do, thanks.

[orion3311]

Hey if you dont mind - i'm working with the exact same setup (and may be able to verify your findings if I can get mine fully working here). How do you specify the backup (MBG2) in ARS? Do I need to have two seperate SIP peer profiles, one for each MBG?

[johnp]

From what I read, I think it's one for each., i.e. 2 sip peers using route lists.

[welshy67]

Hi. Yes, you need two SIP Peer Profiles. We also now use two vMCD's. vMCD1 is primary, vMCD2 is backup. We now have 25 SIP Trunk licenses on each vMCD. MBG1 & MBG2 is in a Cluster with 50 SIP trunks.

On vMCD1:

1. In Network Elements, I have two entries SIP_1 and SIP_2 as Type "Other" and two entries MBG1 and MBG2 as Type "Outbound Proxy".
2. In Trunk Attributes, I have two entries for SIP_1 (Trunk Service number 1) and SIP_2 (Trunk Service number 2).
3. In SIP Peer Profile,  I have two entries for the two Network Elements SIP_1 and SIP_2 with SIP_1 using Trunk Service 1 and SIP_2 using Trunk Service 2.
4. In ARS Routes 10 & 11, I have two entries using SIP Trunk as routing medium and SIP Peer Profiles of SIP_1 and SIP_2.
5. In ARS Routes Lists, I have a List number 10 with 1st choice = 10 (MBG1),  2nd choice = 32 (vMCD2), 3rd choice = 11 (MBG2)

You need to program similar on vMCD2. But in ARS Lists we only have 1st choice to MBG1 and 2nd choice to MBG2. Configuring it this way means we have resiliency on MBG and resiliency on vMCD and resiliency with SIP Trunks. All our devices have a  secondary element of vMCD2.

Friday last week we ported over one of our ISDN numbers to our SIP provider. Shutting down MBG1 WAN link results in incoming calls instantly being re-routed to second SIP circuit and through MBG2 so this is a success. However, outgoing calls do not get routed to MBG2.

I did use Wireshark and the capture file was sent to Mitel who acknowledged that vMCD was working as expected but MBG was not. This has now been escalated to the Mitel MBG Support Team. I will let you know the outcome.

Orion, please let me know how you get on with your setup.