“The odd call drops” of the Mediation Server


I had a very annoying issue lately when an installation of a new gateway resolved in some calls (specifically to US numbers) dropped by the Skype for Business mediation server saying “A call to a PSTN number failed due to non availability of gateways.

The cause, according to the mediation server, was that “All gateways available for this call are marked as down“, and the resolution, surprisingly, was to “Verify that these gateways are up and can respond to calls.”

It seemed funny, because all other calls were successful, I have not exhausted the available PRI channels I had, the gateway didn’t seem to lose connectivity for a split second and SIP options are accepted and replied to on both ends.

Looking further at the Lync Monitoring Reports, I noticed the following:

Reported by Client
12000; reason=”Routes available for this request but no available gateway at this point”

Reported by Server
12000; reason=”Routes available for this request but no available gateway at this point”

I went to the gateway (AudioCodes M1000B) for answers and the logs were showing something very weird;

Every call starts with the same flow:

Flow1The mediation server sends a SIP Invite that will normally be answered immediately by the gateway with a “100 Trying”. This, for me, will close the deal on the “Gateway no responding” – this is a great response if you ask me.
The next phase would be the gateway immediately sending a “PSTN Place call” to the PSTN, hopefully receiving a “Proceeding” instantly from the PSTN, meaning so far we’re fully communicating. So for the time being I’m not really buying what the Lync mediation server is selling.

Looking further at the logs, I see the following strange behaviour:
As I’m still waiting for the PSTN to connect the call (this would normally be a “PSTN Call Alerting” message), I see that the mediation server decided to abandon ship!
This is what I see in the logs:


Within 10 seconds after initiating the call, the mediation server will send a “Cancel” request to the gateway and will terminate the call:

19:25:49.640 [S=660986] [SID:1327438938] INVITE sip:+353857560598@;user=phone SIP/2.0 
19:25:49.729 [S=661027] [SID:1327438938] SIP/2.0 100 Trying 
19:25:49.730 [S=661035] [SID:1327438938] (   lgr_psbrdif)(    656225)   pstn send --> PlaceCall
19:25:50.087 [S=661039] [SID:1327438479] (   lgr_psbrdex)(    656229)   pstn recv <-- CALL_PROCEEDING
19:25:59.020 [S=661055] [SID:1327438938] CANCEL sip:+353857560598@;user=phone SIP/2.0 
19:25:59.052 [S=661061] [SID:1327438938] SIP/2.0 200 OK 
19:25:59.058 [S=661065] [SID:1327438938] SIP/2.0 487 Request Terminated 
19:25:59.079 [S=661074] [SID:1327438938] (   lgr_psbrdif)(    656265)   pstn send --> PSTNDisconnectCall

Clearly, the gateway is available and responding, but the mediation server is somewhat impatient.

Doing a bit of digging around, it appears there’s a setting for this.

The Lync mediation server is expecting a “SIP 180 Ringing” within 10 seconds of initiating the call. Failing to provide that (183 “Session in Progress” can be pushed here but that won’t benefit you if you’re still waiting for that ring…) will cause the mediation server to decide that the call didn’t respond and to mark the gateway as unavailable. A little too harsh? In my opinion yes. This will generate errors (2 per failed call) on your Lync servers and is actually a Lync self-inflicted error.

To work around this issue you can go to:
<Lync or Skype for Business installation folder\Server\Core
and look for a file called “OutboundRouting.exe.config“.
When you open that file, the first configuration line is:
<add key=”FailOverTimeout” value=”10000″/>
This actually determines how long should a call wait for the gateway until it declares it “Offline”. the value is in milliseconds, meaning the default setting is 10 seconds.

This will impact your environment if you have two or more gateways, and it means that it’ll take the mediation servers more than 10 seconds to declare that gateway as offline if it actually fails.

For me to workaround that issue I extended the wait time to 20 seconds, restarted the front-end and mediation services, and all of a sudden no call was dropped.

It appears it took the carrier approx. 12 seconds to generate the “PSTN Call Alerting” command to the gateway. For some reason, only calls to the US (and only with this specific carrier) were affected by this issue.

Now, I can see all my calls working perfectly and the errors have disappeared from the serves too.

11 thoughts on ““The odd call drops” of the Mediation Server

  1. Hello,
    This is just because your configuration on the Audiocodes/IpProfile is not good.
    Lync need to get a 183 Early Media to work and so doesn’t start the timer. If you don’t send a 183 Lync will start the timer and in some case the call is long to proceed so it will be drop by Lync (due to the timer you changed)

  2. soder

    Should warn your readers if any such manual .config file change is done, the next time a cumulative update or patch is installed, it will completely override your customization with the factory default values!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s