Solved Choppy phone calls

Status
Not open for further replies.

MikeyG.IT

Silver Partner
Joined
Oct 2, 2017
Messages
55
Reaction score
2
I have a 3cx deployment that is hosted in the cloud by amazon light sail. I have two remote locations with extensions. One location the call quality is good, but at the other location the call quality is very choppy (in and out audio). The location that is experiencing issues only has 5 extensions. Some additional information: There is a site to site vpn configured between the two remote locations. Both locations have the bandwidth and speeds necessary for VOIP. This system has been deployed for almost a year with no problems and no changes being made to my knowledge and this problem just arose out of the blue. I myself have came to the conclusion that it must be an issue withe ISP but still am not 100%. There could always be an issue with the LAN as well. I have taken a wire shark capture of a call that experienced these issues but I am unsure how to read them and was hoping you guys could help me out in this. The capture can be viewed here [Edited by Administrator]
 
Last edited by a moderator:
So your avatar image says advanced certified which would imply you should know how to read wireshark captures since that's covered in the test....

That being said a packet capture can't really tell you the cause of the choppy audio so much as confirm it and narrow down where to look. Depending on where you are doing the capture you you can narrow down the source. Then once you've done that you can do further testing to pinpoint the issue.
 
I came here for help, not to be told what I should know.... I am NEWLY advanced certified, and am not an expert in reading these captures, hence coming on here for help. I have deployed many systems most of which use voip gateways and consid. I have never needed to use wireshark. So I apologize for not being 100% positive on the capture.
 
And you did get help..

That being said a packet capture can't really tell you the cause of the choppy audio so much as confirm it and narrow down where to look. Depending on where you are doing the capture you you can narrow down the source. Then once you've done that you can do further testing to pinpoint the issue.
 
[Edited by administrator]

@MikeyG.IT have you enabled "PBX delivers audio" under the settings of the EXT in question?

Are able to get the networking equipment rebooted at this site?

Are phones up to date with latest FW?

Is SIP ALG disabled on the router?

what kind of network speeds have they got?
 
Last edited by a moderator:
The capture was not useful. It only showed INVITEs coming from 3CX to two different locations. The capture lacked any RTP details so it is unknown as to which call had an issue as well as which site. The capture did not show the call origination, that being what caused the INVITES to be sent. I assume this is due to the nature of the hosting and that the provider calls are being handled by a different interface than those being sent to the remote locations.

It helps to know more about the situation. So, here are some pointers:
1. Do not post captures that show the public/accessible FQDNs/IPs of the sites. Others may be able to view and use the info to target your system for attacks. Edit the post such that these are identifiable as being internal or external, but without blacking the info out completely so it can no longer be discerned by those that are trying to read them.
2. The good site info is informative as it does imply that the traffic coming in and out of 3CX for the good site is OK, so 3CX is OK. The VPN has no bearing as no VoIP related traffic is seen across the path.
3.
he bandwidth and speeds necessary for VOIP
is a statement that lacks detail to support the claim. a) we do not know if the call issue was heard in both directions and if only one direction, which way? b) we do not know the data needs of the site that is having the issue, which traffic direction may have more utilization and if any QoS is employed. Fiber/Cable or ? Asymmetric or Symmetric speeds? Granted, the voice needs are likely under 500Kbs in either direction, but if someone in the organization does something that involves traffic going up or down, how is the voice traffic protected such that the data traffic doesn't impinge on it by taking all available bandwidth? You have ruled out bandwidth as an issue, but could it be?
4. Are all calls to the problem site affected or is it random calls?
5. What codecs are in use?
6. If a cable modem is in use by your ISP, have you tried to reboot same along with the router?
Cable modems use a mechanism my which they communicate to the CMTS system so they can adjust the RF transmit levels to be optimized for a given site as all may be somewhat different. Similarly, the modem will also adjust the attenuation on the receive side for the same reason. A reboot will cause the training to occur.
7. There are various web sites that offer some form of VoIP suitability testing. Some are free and some paid, but it may pay you to do a search and find one that will simulate the needed number of call that you envision and that will test for some amount of time. Look for latency, jitter and packet loss as testing criteria. This may tell you if the ISP has an issue as these will not get to the LAN side.
 
I have tried the "PBX delivers audio". I have have rebooted all network equipment. I am using codecs that require the least amount of bandwidth. I have ran many voip tests, all of which say the network is suitable. It is only incoming audio that is suffering. There is no active QOS. Both locations pay for symmetrical 10 mgbs but only get around 9 down and about 4 up and have minimal latency. All copper. The problem is at only one location and its every call. Posting the capture before any edits was definitely not a good Idea and I realized the consequences. I was in a panic as my client is basically without phones and I wasn't thinking clearly. I appreciate all the help and advice. The learning never ends! I am getting to the bottom of it and have been working with the ISP and believe that the issue is with them.

Thank you again from all your help.
 
Have you asked the VOIP provider to do a trace and ask for their opinion?

If you enable call recording and make a test call is the choppiness in the recording?

is EXT to EXT audio fine?

If you do *777 from an EXT (echo test) is that audio choppy?
 
Can you/client afford $25 a month? Put an EdgeWater 2900E at the remote site with a public IP on it. It will QoS the voice traffic and allow you to do packet captures remotely. You will also get MOS scores so you can see what the quality looks like. You will definitely see the packet loss and or jitter in the reporting.
 
Sorry for the late reply, and hopefully resolved by now.

The first issue I see is that if both sites are reportedly 10Mbs symmetrical, then why are they showing 9x4? The 9 is not an issue for the test, but the 4 could be (at some point) although the problem being reported is not on the upstream side. Is there some constant off-site back-up running?

I take it that the site with the issue was fine at some point so, the question is what has changed? As the download side seems to be the problem and this is what is the side that most organizations will utilize more so, have their data requirements changed? Have you been able to test knowing the network and WAN load is suppressed such that there should be no congestion? If so, then is does tend to point to the ISP.
 
The first thing that jumps out to me is "site to site VPN" to the cloud instance. I'm no fan of that. You cannot prioritize the RTP inside of a VPN tunnel. You're probably getting creamed there.

Install the 3CX SBC and connect that to your instance instead ... you can prioritize that a bit.

Also, someone suggested an edgewater edgemarc ... we LOVE those here.

Edgewater + 3CX SBC and you should be good (or at least "best positioned"). Only with the SBC you won't get MOS. You're only going to get that if you drop the 3CX SBC and properly config the EM's Voip, Survivability, and Traffic shaper.

Just hope it's not a problem with your ISP connecting to that cloud endpoint.

Another idea ... run WinMTR and see what your latency looks like when the calls go to heck
 
It was an ISP issue. The problem has been taken care of. Thank you everyone for your help!
 
Glad to see the issue has been resolved and thank you for updating the thread with your solution
 
Status
Not open for further replies.

Latest Posts

Forum statistics

Threads
111,901
Messages
589,636
Members
164,768
Latest member
Eagle Man