Sip Show Channelstats Reliable?

Home » Asterisk Users » Sip Show Channelstats Reliable?
Asterisk Users 6 Comments

I am seeing lots of lost packets when running the command sip show channelstats at the CLI. There are issues across multiple Asterisk servers I am trying to diagnose but everything I read seems to point to this command being pretty unreliable. Can I trust the info this command shows?
I am showing lots of lost packets in sip show channelstats but I can’t see any packet loss when pinging the same IP’s to/from. Since I don’t 100% control the network my gear is on, I need something outside of Asterisk to show the network engineer to convince here and myself that there are network issues. All I have is the loss that’s shown from this command with no real network stats to back it up. Is there a magic command in CentOS anyone can recommend to diagnose and match up the issues shown in Asterisk using this command?
Moving gear around on the network changes the info Asterisk shows a LOT. For example, if I point traffic to the main physical gateway I get loss to a particular customer’s IP (their PBX), if I move it to another place on the network (as a VM) their IP is good and other customers IP’s start showing loss using the channelstats info. Driving me freakin’ crazy. It does appear there are network issues causing my troubles but I can’t get help if I can’t point to some hard and fast issues outside of Asterisk. The only thing I have right now is collissions showing on one of a few of our pfSense devices but they are virtual running on XenServer, still this would indicate a problem in my opinion. Thanks in advance for any assistance on this issue. Stepping back from the ledge now LOL

6 thoughts on - Sip Show Channelstats Reliable?

  • Additional info:
    At the moment I am running 1.8.x but the other day I was getting the same results on 11.x Here is a sample from show channelstats. I do think this command is showing that there is trouble between specific IP’s and my Asterisk box but I don’t know if the numbers are accurate and reliable.

    Peer
    Call ID
    Duration
    Recv: Pack
    Lost
    ( %)
    Jitter
    Send: Pack
    Lost
    (
    %)
    Jitter

    x.x.x.x
    5531341d06b
    00:07:42
    0000023123
    0000063836
    (73.41%)
    0.0000
    0000023102
    0000000000
    (
    0.00%)
    0.0007

    Peer IP changed to protect the innocent 🙂

    From: tjrlist@live.com To: asterisk-users@lists.digium.com Date: Mon, 19 Jan 2015 12:17:25 -0600
    Subject: [asterisk-users] sip show channelstats reliable?

    I am seeing lots of lost packets when running the command sip show channelstats at the CLI. There are issues across multiple Asterisk servers I am trying to diagnose but everything I read seems to point to this command being pretty unreliable. Can I trust the info this command shows?
    I am showing lots of lost packets in sip show channelstats but I can’t see any packet loss when pinging the same IP’s to/from. Since I don’t 100% control the network my gear is on, I need something outside of Asterisk to show the network engineer to convince here and myself that there are network issues. All I have is the loss that’s shown from this command with no real network stats to back it up. Is there a magic command in CentOS anyone can recommend to diagnose and match up the issues shown in Asterisk using this command?
    Moving gear around on the network changes the info Asterisk shows a LOT. For example, if I point traffic to the main physical gateway I get loss to a particular customer’s IP (their PBX), if I move it to another place on the network (as a VM) their IP is good and other customers IP’s start showing loss using the channelstats info. Driving me freakin’ crazy. It does appear there are network issues causing my troubles but I can’t get help if I can’t point to some hard and fast issues outside of Asterisk. The only thing I have right now is collissions showing on one of a few of our pfSense devices but they are virtual running on XenServer, still this would indicate a problem in my opinion. Thanks in advance for any assistance on this issue. Stepping back from the ledge now LOL

  • I’ve seen something similar with Adtran SIP gateways. When a re-invite happens the Adtran gets all confused about call stats and marks the pre-reinvite leg of the call as losing large numbers of packets. BTW, IIRC reinvites happen when a codec changes or the channel switches to T.38.

    Also Adtran SIP gateways appear not to support OPTIONS packets when running in SIP proxy mode, which is very annoying. At some point I’ll try and arrange a slugfest between Digium and Adtran and they can figure out why it doesn’t work.

    From: asterisk-users-bounces@lists.digium.com [mailto:asterisk-users-bounces@lists.digium.com] On Behalf Of Todd R. Sent: Monday, January 19, 2015 1:45 PM
    To: Asterisk-Users List Subject: Re: [asterisk-users] sip show channelstats reliable?

    Additional info:

    At the moment I am running 1.8.x but the other day I was getting the same results on 11.x

    Here is a sample from show channelstats. I do think this command is showing that there is trouble between specific IP’s and my Asterisk box but I don’t know if the numbers are accurate and reliable.

    Peer

    Call ID

    Duration

    Recv: Pack

    Lost

    ( %)

    Jitter

    Send: Pack

    Lost

    (

    %)

    Jitter

    x.x.x.x

    5531341d06b

    00:07:42

    0000023123

    0000063836

    (73.41%)

    0.0000

    0000023102

    0000000000

    (

    0.00%)

    0.0007

    Peer IP changed to protect the innocent 🙂

  • I would recommend capturing traffic outside your Asterisk server with Wireshark, then running the Telephony/Rtp/Analysize Streams option to determine if you have packet loss at that point in the network.

  • On Tue, Jan 20, 2015 at 12:43 AM, Scott Griepentrog
    You can find out the data loss outside of Asterisk by using tcpdump and tshark(wireshark)

    1. Capture output of Asterisk SIP channels in a log file ax_log_yyyymmdd

    $while :; do date; asterisk -rnx ‘sip show channelstats’; sleep 5 ; done

    2. Capture tcpdump traffic on the asterisk server:

    $tcpdump -nq -s 0 -i eth0 -G3600 -w eth_sip_traffic-%F-%H-%M-%S.pcap port
    5060 or port 5061
    [this saves the all the ethernet traffic of ports 5060 & 5061 in the pcap file for every hour(-G 3600) ]

    3. Once you can see the data loss in the ax_log_yyyymmdd, check for the same time in the eth_sip_traffic.pcap

    Analyze the eth_sip_traffic.pcap

    $tshark -t ad -r eth_sip_traffic.pcap |grep sip_client_ip | less
    [ -t ad: is for time format, -r :is for input file]

    1034847 2000-01-03 22:08:10.239661 sip_client_ip -> asterisk_server_ip RTP PT=ITU-T G.711 PCMA, SSRC=0x488EDB49, Seq=314, Time=50240
    1036396 2000-01-03 22:08:11.647404 sip_client_ip -> asterisk_server_ip RTP PT=ITU-T G.711 PCMA, SSRC=0x488EDB49, Seq=383, Time=61280
    1036401 2000-01-03 22:08:11.647560 sip_client_ip -> asterisk_server_ip RTP PT=ITU-T G.711 PCMA, SSRC=0x488EDB49, Seq=384, Time=61440

    You can find the if the packets loss is happening, with the missing sequence numbers.

    PS: I think any loss greater than 3%, will deteriorate the call quality.

  • Is it possible that this kind of packet loss in sip channels can cause High load on the server?