SOLVED A few issues - Audio drop incoming/Call Drops/Multi Registration so on

[cengbrecht]

Hey Guys!
Still new here, and have posted before, but I need some insights, and yes, last time it was something that was my fault.

Anyway, this time I am not sure whats up.

My office, and my church office are both running on my server.
My office can make many hour calls, no issues, but the church here has issues with outgoing, and incoming calls where the audio from the far end sometimes goes quiet.
I have checked the NAT, I have checked the load. So on...
The Specs :
Dual thread Intel hosted on Digital Ocean Toronto. (From Webmin) Intel(R) Xeon(R) CPU E5-2650 v4 @ 2.20GHz, 2 cores
2 GB ram.
30GB SSD backed drive.
1Gbps connection to the web. 1TB of traffic.

Church office:
30/30 connection, with 5-15ms of jitter, with under 7-14 to the server.
Unifi router - Unifi Switch - pfSense in passive mode for online filtering of DNS nothing else - Client computers and Phones, Computers plugged into Phones.

Phones VVX410 all new, no known issues with the batch.


Calls coming in and going out have on a random tendancy to drop only the incoming audio, and with no known outgoing audio issues. ...

Issue the second:
Some calls get dropped, much the same info above, please inquire if you have need of information.
To my knowlegde, these happen only in a limited case, and usually within the first seconds of a call.

Issue the third:
Phones De-register, but I have noticed that some had a negative expsecs value, after de-registering all the phones including the negative value ones, everything seems to be online again...


Please help.
Signed,
A guy with lots of hair, who doesn't want to be bald, but shaves his head...

 

[DigitalDaz]

I do not know if those Polycoms have an rport setting on, can you take a look and see and if there is, enable it.

 

[cengbrecht]

What does the rport do? I have looked into this, and aparently H.323 (Similar) can cause this issue, I have tried it enabled, and disabled.
I have tried SIP_ALG, and many other things... so far, it seems NO change has helped this audio issue, which seems to be LAN side only, where the caller externally hears all audio, and the receiver internally should be receiving it, but looses it locally.

 

[Stelios]

How many Gateways you have?

Make sure that in sip status you have the same number of Gateways.

About the Audio make sure that you all range of RTP ports (Audio Ports) are open.

 

[cengbrecht]

There is only one Gateway, we use a local Internet provider that does wireless bridge, the link is not delayed, and they do not filter ports or anything.
All RTP ports on the server side are open, it is not an issue where we hear no audio from the remote, its intermittent, and comes back.

On the other note, the phones have stopped de-registering, it seems that it was a programming glitch, meaning they register many times in a few minutes due to re-provisioning so many times, as the phones are manually programmed by local tftp server so I can do updates, as Fusion does not seem to support VVX410's properly, it shows the right firmware, but doesn't do BLF correctly, and other features.
Also, I don't seem to be able to set the password through the web interface via Fusion either...

 

[cengbrecht]

I have disabled all the CONNTRACK MODULES on the Unifi network, most of these are enabled on my office network where there are no issues, but this local network is a LOT larger, so there could be something there.
The registrations seem to be good now, and I have some users pestering me. But the audio incoming still goes silent for short periods, and I am clueless as to why.

 

[cengbrecht]

This may also be a subnet issue, though I should hope not.
The subnet is a /23, and almost all of the phones are in the root range under a 24, but it does extend up one.
I have disabled the CONNTRACK and just done a full reboot of the wider network.

Also note that the calls locally also have audio drop issues on the receiving end, locally I still heard noise.
I believe this is from the switch to the phones, not PBX to local network.

 

[cengbrecht]

Also, please note that Polycom VVX410's do not have a jitter buffer as of 5.x.x.x

 

[cengbrecht]

So I have done a bit of capture, and it looks like the phone is sending all the audio, but the switch is not...
Any suggestions?

 

[Stelios]

What about codecs?
Do you have any issue if you make calls between extensions or the issues take place only from inbound calls (from a voip provider)?
Are you sure that all parties support the negotiations codecs ?

 

[cengbrecht]

There does not seem to be any codec issues, as I have mentioned above, the calls all have audio, and the phones receive the calls, with no issues, but the audio drops off intermittently.
The audio from ext to ext also can drop out intermittently bidirectionally, but the call stays active.

 

[cengbrecht]

After checking with Ubiquity, and testing out a few settings, it doesn't seem that the setting changes have helped the audio issues.
Still looking into it.

 

[bcmike]

So if I understand you correctly the users at your church connect to a Fusion PBX instance hosted at Digital Ocean. You have intermittent audio problems even between locals and you only have one gateway device.

I would start with simple things and do a ping plot or an MTR from your church network to the Digital Ocean instance. You need to see what the connection is doing over time and try to correlate that with your audio issues. My guess is that you're seeing intermittent packet loss and or congestion to the public network. You may even have a specific issue with the route to Digital Ocean.

I'd try and eliminate any basic connectivity issues first before you chase your tail with router/switch settings, etc. The fact that they register and set up a call properly eliminates a lot of that NAT stuff (unless the mapping gets nuked by a timer).

Also who is your ISP? Some are better than others for VoIP.

 

[cengbrecht]

Normally I would agree with you, but there is no congestion on the line, and live monitoring of the network shows no overload.
In all of my testing, only the local lines are affected, the recordings have all audio, and the wan side does receive the packets, but the local switch or router seem to drop them or something.
I am in communication with Ubiquity about this issue, as it seems to have been more widespread.

As for the provider issue, the WISP we use has excellent service, and we have had no issues on his network yet.
On that note, we are using Shaw behind that.

 

[bcmike]

Are you using Terrago? Check to see that the radio is capable of doing the appropriate packets per second for VoIP. Also Shaws Docsis network is not the best choice for VoIP (although I know you cant just drop your ISP on a whim).

On your Ubiquiti try disabling hardware off loading if it is enabled, also any SIP ALG or SIP contrack modules.

I'm curious to know how you're monitoring for network congestion.

 

[cengbrecht]

We are using all Ubiquity network hardware from Shaw term to our network.
No terrago stuff, are you in Alberta? Edgar High Speed, I have talked to them, and they have monitored the outside traffic, and noted no failures, the packets all get to the router. I have no dropped packets on the Ubiquity router side internally. I have not monitored the switch closely enough to know though for the switch, but the phone side received no packets when the call audio stops; It was however, still sending.

 

[bcmike]

I hate to sound like a broken record, but having the ISP test your connection is unreliable as they won't test continually over time. Also the stats on the ubiquity hardware won't tell you if there's a problem on a hop in between yourself and Digital Ocean.

Take my advice and download ping plotter: https://www.pingplotter.com/ and enable ICMP on your digital ocean box. Run ping plotter for a at least 48 hours and look to see what the results are. It will tell you if there's any packet loss and at what hop.Make sure to set # of times to sample as unlimited, your trace interval to 1 second and samples to 5. If you need help dissecting the plot let me know but its fairly straight forward.

From the fusion pbx itself you can run MTR (Matts Trace Route) the other way but I like ping plotter better as its easier to read.

If your plots are clean over time, then start looking at hardware, etc..

 

[bcmike]

Sorry forgot to mention, I'm in Vancouver.

 

[cengbrecht]

This particular ISP has a SLA with us, and monitors the link 24/7/365.
For the sake of argument, I could connect a phone direct and see if there are any issues.

 

[bcmike]

But how does the ISP monitor the link? From their gateway to your end point or beyond? If its a ho in between how would they know? At what interval? Packet loss can occur for 5 seconds and then go away without being detected by a system that monitors in 20 second intervals, however that 5 seconds is a long piece of dead air for a phone call. The ISP SLA probably covers four or five nines of up time therefore contratually they're worried about minutes and hours of downtime, not seconds of downtime.

You need to setup your own test within your parameters to be sure.