unreliable SIP calls

Discussion in 'General' started by jmsaul, Mar 2, 2011.

  1. jmsaul

    Joined:
    Mar 2, 2011
    Messages:
    5
    Likes Received:
    0
    I hope someone has someone can help with an ongoing issue we have been experiencing over the past months. I have an Elastix box configured and setup with Polycom 331 soundpoint phones (both on the LAN and in remote office over the internet). My intent is to use the box purely with a SIP provider for calling in/out of the office. Sadly we can only achieve about 85% stability with call quality (garble) or drop calls occurring the remainder of the time and in an unpredictable manner. I have tested with 3 different SIP providers using both sip and iax2 connections.

    What is interesting is that we can have simultaneous calls on the system and whereas one might become garbled the others are okay. The effected calls always seem to involve the SIP line while direct phone-to-phone calls never seem to have an issue (even if one phone is remote). In fact- I have even had a situation where within a conference room with 3 outside callers dialed in through the SIP provider one caller becomes garbled and all the others remain fine. It seems like if it were simply a network problem everyone would be effected.

    I am connected via. Verizon FIOS 25MBs up / 25MBs down using there standard issue westell modem. I have -- as best I can tell-- disabled SPI and any other firewall issues on the modem. Most calls are ulaw but g729 (free codec) yields similar results. Setting any level of jitter buffering seems to make things much worse. And as I say the system works great 85% of the time.

    I am about to abandon SIP and order a digium card and install a batch of analog lines. I hate to do it... but this has got me stumped.

    Thanks for ANY suggestions!
     
  2. Lee Sharp

    Joined:
    Sep 28, 2010
    Messages:
    332
    Likes Received:
    0
    What is going on with your network at the time? Are you running any QOS? Have you logged any of the calls?
     
  3. jmsaul

    Joined:
    Mar 2, 2011
    Messages:
    5
    Likes Received:
    0
    I am running QOS at the router. Wireshark inspection does show packets are QOS tagged. I have logged calls where there is audio distortion. Catching a drop is more difficult. Generally the network stats look good and further more-- one call maybe afflicted while other simultaneous calls are fine. At the same time the issue appears independent on the number of concurrent calls. So network and bandwidth seem okay. I had elastix paid support take a look and they found nothing conclusive or helpful.
     
  4. hinzinho

    Joined:
    Sep 18, 2009
    Messages:
    461
    Likes Received:
    0
    If your SIP provider is going over the internet, QoS will not help. The best guarantee to have good quality calls is have a point-to-point cable to your SIP provider. That's my 2cents.
     
  5. fmvillares

    Joined:
    Sep 8, 2007
    Messages:
    1,785
    Likes Received:
    0
    as probably u are new to voip with total security you have a bad install or something wrong configured in your systems...what kind of switches? cpu ? ram?elastix version? you are using vlans? pc connected to the phones also?
    Sip trunks are very reliable if configured correctly and by professionals.
     
  6. Lee Sharp

    Joined:
    Sep 28, 2010
    Messages:
    332
    Likes Received:
    0
    Do a traceroute between you and the sip provider. Post it with your IP address munged...
     
  7. jmsaul

    Joined:
    Mar 2, 2011
    Messages:
    5
    Likes Received:
    0
    Thanks for the responses!

    I am running Elastix 2.0.0-36 on a Dell PowerEdge quad core intel 2.675 Ghz server with 4 Gigs of RAM. The server never registers any significant load. It is connected directly to the FIOS router at 100mbs. The router has a 4 port switch which is then connected to a netgear Gigabit switch to which each LAN phone is attached. Most phones then do have a cable running to a computer station. There are no vlans or other network division or subnets.

    The configuration is pretty much as suggested by the SIP provider. I have several disabled trunks from previous attempts to use alternate providers. It's possible some stray configuration changes could be hanging about but nothing major. The setup was reviewed by the Elastix support guy and looked good to him. I could do a re-install if anyone thinks that's a reasonable next move.

    Network analysis shows a 25up/25down connection to FIOS with lesser speeds maintained out into the nets. Here is a traceroute to my provider. FIOS seems to terminate in Atlanta even though I have an option for a more local (tampa) SIP POP-- that shows higher latency by ping so I use the atlanta pop.

    Here is a traceroute from my box to the POP.

    traceroute to atlanta.voip.ms (174.34.146.162), 30 hops max, 40 byte packets
    1 myrouter.home (X.X.X.X) 0.996 ms 1.199 ms 1.392 ms
    2 L300.TAMPFL-VFTTP-101.verizon-gni.net (96.243.202.1) 16.643 ms 16.807 ms 16.937 ms
    3 G10-3-1101.TAMPFL-LCR-01.verizon-gni.net (130.81.129.64) 17.044 ms 17.200 ms 17.345 ms
    4 so-7-2-0-0.TPA01-BB-RTR1.verizon-gni.net (130.81.28.233) 17.566 ms 17.722 ms 17.872 ms
    5 so-7-3-0-0.ATL01-BB-RTR1.verizon-gni.net (130.81.19.30) 24.240 ms 24.392 ms 26.366 ms
    6 0.xe-7-1-0.BR3.ATL4.ALTER.NET (152.63.80.73) 81.775 ms 44.966 ms 45.034 ms
    7 67.111.23.189.ptr.us.xo.net (67.111.23.189) 24.589 ms 24.360 ms 24.503 ms
    8 vb2000d2.rar3.atlanta-ga.us.xo.net (207.88.13.158) 26.521 ms 24.359 ms 24.508 ms
    9 te3-0-0.cvr1.atlanta6-ga.us.xo.net (207.88.14.2) 41.857 ms 41.829 ms 42.250 ms
    10 edge1.atl.ubiquityservers.com (207.88.188.30) 42.219 ms 46.839 ms 46.967 ms
    11 174.34.146.162.rdns.ubiquityservers.com (174.34.146.162) 39.312 ms 39.549 ms 39.353 ms
     
  8. jmsaul

    Joined:
    Mar 2, 2011
    Messages:
    5
    Likes Received:
    0
    Just to be certain--


    It was suggested at one time that this could be a timing source problem. I have no analog lines and hence no analog interface or other specific timing source. My understanding is that the timing source is therefor handled by asterisk and only maybe relevant to the conference room (which works fine but for the stated global SIP problems).
     
  9. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    To be technical, the timing source in the absence of proper hardware is provided by dahdi (dahdi_dummy is the driver) not by Asterisk. so just for completeness, make sure:

    dahdi_test -vv

    consistently returns "three and a half nines" or better.

    A couple of comments and a suggestion:

    Your line 2 in the traceroute shows an unexpected lag for a FIOS connection. ICMP will defer to other traffic, so run the same test in the dead of night also :)

    You do not state what phones you are using but not all support QOS on the pass through port so a noisy user might well be flooding the shared LAN , I suggest you implement a VLAN if supported on the phone (traditionally tagged on 512)) for your voip traffic and thus isolate the PC stuff , locally manage the VLAN too as to TOS/QOS, the defaults are often wildly off if set up wrong, only award 20 Meg to the shared bandwidth if you have a reliable 25 , or the "buckets can get full", don't rely on the provided gateway for QOS, it almost certainly is not up to it in a heavily used network.

    To debug from Asterisk:

    rtp set debug IP <a chosen IP, local or otherwise>

    and

    tailf /var/log/asterisk/full

    to keep your blood pressure down, the ins and outs and the time-stamps need to be impeccable ordered, so watch/explore what happens when "some-one gets garbled". One is likely to slip but it will provide ythat essential clue.
     
  10. jmsaul

    Joined:
    Mar 2, 2011
    Messages:
    5
    Likes Received:
    0
    THe phones are polycom soundpoint 331. I don't have a vlan setup and I am not sure if the soundpoints support that. I guess I keep coming back to the fact that phone to phone calls work fine (even out through the internet) but calls over the sip provider are unstable. To me that suggests the network is not creating a major problem. Maybe that's erroneous logic. I revisit the QOS capabilities of the router. They are limited but perhaps there's something going on at that level that is not as expected.

    the dahdi_test yields: --- Results after 28 passes ---
    Best: 99.999 -- Worst: 99.990 -- Average: 99.996267, Difference: 99.996531
     
  11. dicko

    Joined:
    Oct 24, 2008
    Messages:
    4,099
    Likes Received:
    0
    Yes Polycoms suppor vlans. And your system also does, look into

    yum install vconfig

    and then

    man vconfig

    But your system seems really quite adequate timing and bandwidth wise (provided you have eliminate all the porn and music that your users might be downloading), you will now need to do the rtp debug/wireshark thingy, I would suspect your provider, all other network things being equal, try Vitelity or Teliax as reliable touchstones in NANP-land. before you give up.

    Anecdotally , and noting your location, I have had several run-ins with RoadRunner as they have arbitrarily blocked rtp connections for several seconds every five minutes or so, they deny it of course but, my logs indicated otherwise . . ., my pragmatic solution was to cancel service, you can't beat the big boys after all. I believe there is a class action against them for this issue, I expect my $1.23 sometime in the next twenty years.

    dicko
     
  12. fmvillares

    Joined:
    Sep 8, 2007
    Messages:
    1,785
    Likes Received:
    0
    Re: Re:unreliable SIP calls

    agreed with dicko look into the filtering your provider maybe could be doing
     

Share This Page