Dismiss Notice
We would like to remind you that we’re updating our login process for all 3CX forums whereby you will be able to login with the same credentials you use for the Partner or Customer Portal. Click here to read more.

Solved Extensions constantly register/unregister

Discussion in '3CX Phone System - General' started by scotru, Dec 4, 2017.

Thread Status:
Not open for further replies.
  1. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Greetings,

    I am working on a new 3cx installation. I'm having a problem where all extensions are constantly registering/unregistering (this includes system extensions). I'll start the server up--everything will look good. I was able to briefly have a conversation between 2 extensions. But when I tried to call back between them, one of them had unregistered. A moment later both were unregistered. My system extensions are doing the same thing, they constantly fluctuate between registered and unregistered--not all at once, but one at a time. Each refresh of the screen shows different extensions (phone and system) registered. Essentially my extension status panel is blinking like xmas lights.

    3CX is running on a Hyper-V VM. There's a dedicated NIC in the host for the PHONE network (which also gets it's own port on the router--a Ubiquiti ER-Lite). There's a subnet setup for the phone network and a dedicated switch to connect it all together. No problems showing with network connectivity (I can ping between phones and 3CX VM continously with no dropped packets). I can make outgoing calls from an extension while it is registered. Generally if I go to the provisioning page and hit save, the extension comes registered for a little bit (couple minutes). Anytime I try to receive an outgoing call to an incoming extension, the extension immediately goes unregistered and the call goes to 3cx voicemail.

    I've tried deleting the VM and starting over with a new installation multiple times and the same issues persist. The phones are Yealink T46S and T42S's (all extensions experience the issue). They were auto provisioned by 3cx. Firewall test passes. Anyone have any ideas? I'm beating my head against this one.

    Thanks!
     
    #1 scotru, Dec 4, 2017
    Last edited: Dec 4, 2017
  2. lneblett

    lneblett Well-Known Member

    Joined:
    Sep 7, 2010
    Messages:
    2,086
    Likes Received:
    64
    I have not experienced this, but as you have indicated that the issue is also impacting system extensions, then it seems that something more problematic is at hand with the install itself. You might check the Windows system logs and see if that reveals anything. On the surface, I can only guess that it has to do with the VM instance.
     
  3. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Sorry, another piece of information I should have included, this is on Debian. So far, I haven't been able to find anything noteworthy in the system logs.
     
  4. nitrox

    Joined:
    Sep 10, 2016
    Messages:
    95
    Likes Received:
    11
    Did you try to install 3cx 15.5 SP2? I mean the new iso version with debian 9. Could you tell us the hardware setting of the VM? CPU/RAM/HD/NUMBER_OF_EXTENTION/NUMBER_OF_BLF/NUMBER_OF_SIP_TRUNK etc etc. Did you notice any flapping of the sip trunk as well?
     
  5. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Thanks for your response.

    That's correct--latest version. I originally downloaded the PIAF version from pbxinaflash, but it seems to be identical to the 3cx download. It is Debian/GNU Linux 9. Here's the VM settings (current and some older things I've tried):

    Generation 1
    RAM: 4GB (Originally set to dynamic but now static as I thought that might be a problem)
    Processor: I've tried both 1 and 2, currently 2
    HDD: 127gb IDE emulated VHDX
    Network Adapter: I tried disabling VMQ here under Hardware Acceleration and also through PowerShell based on this: https://support.microsoft.com/en-us...work-connectivity-when-you-use-broadcom-netxt article that was linked to from 3CX configuration page (although I didn't notice any network issues--pings are consistent). I do have a Broadcom NetXtreme based network card in the host. I also have a static MAC address set on my network interface.

    I have 5 extensions right now, plus system extensions, although I started noticing problems with even the first extension I set up going on and off. I have a single SIP trunk (Twilio) and I haven't noticed any problems with it (though I'm using IP based authentication so I don't think it registers).

    The fact that the problem effects system extensions makes me think PBX system, but I've done several reinstalls (without restoring backup, just resetting up), tried different VM settings, etc.. This makes me suspect network--but no issues with pings or network connectivity that I can tell.

    NOTE: I don't think this is relevant, but the very first time I tried to set things up, I did accidentally assign an IP I'd already used in the network to the PBX server, but this was quickly resolved (by moving the other device to a different IP) and I've done a fresh install of 3CX since then.

    I also have a very similar deployment working fine elsewhere (also running in an identically configured VM). But it is running an older ISO with Debian 8.
     
  6. nitrox

    Joined:
    Sep 10, 2016
    Messages:
    95
    Likes Received:
    11
    ok. If it is a dns problem? Look at the other configuration and see if you use different dns. Sometimes the problem it's easier than we think :) or I hope so. It would be nice to see a trace during the deregistration. If you can trace in middle probably we can see something.
     
  7. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    DNS is different between the two sites (both use internal Windows servers for DNS that forward unknown requests). nslookup resolves common sites, etc.. on the 3cx server all right. The network does have DNS based filtering in place (it's a school). So that is a thought--but I'd be surprised if anything that 3cx was trying to resolve would be blocked. I'll see if I can temporarily disable the filtering for testing.

    A trace shouldn't be hard to get. I'll go out tomorrow and mirror a port on the switch and do a trace, scrub it, and upload here.
     
  8. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Sorry this took so long. Here's a trace (run through TraceWranger to randomize IPs/MACs/etc...). During this time I made an edit to the extension on the server and saved it (which always makes the extension show registered for a moment). Saw the extension go to unregistered in the view, and made a test call from outside and had it go to voicemail instead of to the extension.

    I appreciate any ideas here.
     

    Attached Files:

  9. lneblett

    lneblett Well-Known Member

    Joined:
    Sep 7, 2010
    Messages:
    2,086
    Likes Received:
    64
    The capture showed nothing that it could identify as SIP,. it did however show what might be some pertinent data in that the phone and the 3CX server may be on different subnets. I say may as it appears that you have a limit set on what the capturing system is allowing. As a result, the capture is truncated and does not show the entire entry. I cannot identify which IP is which device nor what the message was, but at least it is showing the IP and port (5060) in the source and destination as well as:
    [Packet size limited during capture: SIP truncated]

    You need to expand what the capture will allow, this may be under payload settings.
     
  10. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Sorry, my attempt to anonymize the packets with TraceWrangler (default setings) seems to have been overzealous. I've scaled back the settings now and I think the relevant packets are no longer being "sanitized". You should be able to see the SIP traffic between the phone (192.168.25.100) and the 3cx server (192.168.25.10) as well as between the 3cx server and our trunk provider. I've replaced our outside IP with 1.2.3.4. But otherwise this trace should be pretty close to the original.
     

    Attached Files:

  11. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Also, does anyone know if the Debian 8 version of the 3cx ISO is still available somewhere? I'd like to try moving back to Debian 8 and see if that helps.
     
  12. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    Anyone have any other ideas on this?
     
  13. YiannisH_3CX

    YiannisH_3CX Support Team
    Staff Member 3CX Support

    Joined:
    May 10, 2016
    Messages:
    7,326
    Likes Received:
    532
    Hello @scotru

    You could have an issue with time no being accurate on your installation that might be causing extensions to unregister since you are using Debian and HyperV. Try installing an NTP client on your Debian 9 installation and see if the issue persists.

    Code:
    apt install ntp
     
    scotru likes this.
  14. rks

    rks

    Joined:
    Aug 6, 2016
    Messages:
    15
    Likes Received:
    1
    Hi please dont waste anymore time on this. It's a bug but 3CX support hasn't been much help on this issue. It appears to be with the debian 9 iso. Use the debian 8 iso and you should be good to go.
     
    scotru likes this.
  15. nitrox

    Joined:
    Sep 10, 2016
    Messages:
    95
    Likes Received:
    11
    I have the same problem on debian 8 in one installation with snom phones. I am trying to understand the issue. Why are you so sure that is a 3cx bug?

    PBX: 15.5.6354.2
    Snom phone 710 firmware: 8.9.3.66
     
    #15 nitrox, Dec 29, 2017
    Last edited: Dec 29, 2017
    MpDay likes this.
  16. rks

    rks

    Joined:
    Aug 6, 2016
    Messages:
    15
    Likes Received:
    1
    We've had this issue with several debian 9 installs on our Hyper V environment. When we use the 8 iso or Windows version no issues.
     
  17. YiannisH_3CX

    YiannisH_3CX Support Team
    Staff Member 3CX Support

    Joined:
    May 10, 2016
    Messages:
    7,326
    Likes Received:
    532
    Have you tried installing an NTP server on your Debian 9 installations? Did the issue persist? This could be a time issue.
     
    scotru likes this.
  18. rks

    rks

    Joined:
    Aug 6, 2016
    Messages:
    15
    Likes Received:
    1
    I haven't tried installing the NTP server I assumed this was something that was installed by default is that not the case? I'll deploy a test vm and report back with my findings.
     
  19. scotru

    Joined:
    Dec 4, 2017
    Messages:
    9
    Likes Received:
    0
    @YiannisH Thanks! I can confirm that installing NTP resolved the issue for me.
    Another clue that this may have been time related was that the performance graph in the Dashboard was sometimes drawing loops or circles, rather than being a continuous function.

    Might I suggest including NTP in the next build of the ISO image or adding it's installation to the documentation?

    Thanks!
     
  20. NickD_3CX

    NickD_3CX Support Team
    Staff Member 3CX Support

    Joined:
    Jun 2, 2014
    Messages:
    1,375
    Likes Received:
    84
    To explain what is happening as some people rush to say "3CX bug", starting with Debian 9 Stretch, there is a time sync service in systemd, theoretically making the "ntp" package unnecessary. Unfortunately though not all Virtualization platforms work in the same way. Hyper-V is unable to allow the systemd time sync service to update the time correctly from the host, whereas the "ntp" package is.

    This is the reason why in the "Installing 3CX Phone System on a Hyper-V VM" we have a section specifically stating this: https://www.3cx.com/docs/installing-microsoft-hyper-v/#h.t18y0ka76ewm

    Addressing the suggestion that we include the ntp-client package in our Debian 9 ISO, we have tested installing it even on VM platforms that don't have this problem and to our knowledge it causes no issues. We did though choose to trust the decision of the people that made Debian 9 Stretch to not include it in the vanilla ISO and rely on the built-in systemd time sync function. For this reason we just added this as an extra step in the 3CX-HyperV guide.
    Other supported/recommended VM Platforms seem to be unaffected.
     
    nitrox likes this.
Thread Status:
Not open for further replies.