SV9100 - SIP trunk disconnects/issues

[Frank_And_Beans]

Hello,

I'm posting this as I've exhausted almost every other avenue including our PBX vendor, Edge (SIP vendor), and NEC but still have not been able to resolve the issue. For the past 5-6 months, we have had issues where the SIP trunks are down/disconnected. This manifested itself initially as customers complaining about our phones being busy. Then some of our employees complained that they could not make outgoing calls. I initially thought maybe we had more demand than SIP supply. I contacted Edge and they provided usage reports. Those reports showed that at no point have we ever had all 8 SIP trunks in use. We have a static IP and this is set in MB 10-12. The PBX vendor had me try switching the PBX to use IP instead of DNS, that didn't help. They also had the SIP provider change the registration process but that didn't help.

Once those settings were ruled out, I engaged our PBX vendor to see if it was something related to our firewall. We are using a USG4 Pro which is made by Ubiquiti. I do not believe this is a firewall issue as I manage another site that has a SV9100 and a Ubiquiti firewall as well and they do not have these issues. Nonetheless, I am currently running Wireshark on both the LAN and WAN side but after providing several logs to both our vendor and NEC, nothing has stood out as the issue.

If you log into the PBX and go to the System Alarm page, you can see each time the SIP trunk goes down. What is odd is that when it goes down, it takes exactly 30 minutes before reconnecting. I've attached an image showing this pattern. I've also found that if I reboot the SV9100 (System Initialization > 2nd Initialization), the SIP trunks come back online as soon as the PBX is has completed its reboot. This avoids/bypasses the 30 minute duration/downtime. Seems to me that it has something to do with the SIP provider or the PBX itself.

Any ideas/shots in the dark are greatly appreciated! Thank you.

 

[OzzieGeorge]

At what point is the outgoing call blocked? can you get a line and dial the number? Reason I ask is that here in OZ most providers will block outgoing if the CLI is not active on the service (to prevent spoofing) could this be an issue?

 

[Frank_And_Beans]

Hi OzzieGeorge,

When you say, "at what point is the outgoing call blocked?", do you mean how far into entering digits does the system start making the busy signal? Apologies but my PBX terminology is atrocious. If so, it is as soon as the PBX attempts the call, then it is either the 4th or 7th digit depending if the area code is being entered or not. For example XXX-XXXX, then the 4th digit, if XXX-XXX-XXXX, then the 7th digit. I guess another way to say it is, when the system attempts to grab a trunk or recognizes the call is not internal.

Thank you!

 

[OzzieGeorge]

Ok so have you tried setting a line key on a digital phone and then trying to select that and then dialling the number (witout the line access code as in theory you already have a line). When does it reject the call then, can you actually get a line when pressing the line key. The reason I ask is that you may not have added the new SIP trunks to the outgoing group or something equally simple to overlook. With a line key, you bypass such things and if you can't get the line, it pretty much means you aren't registering.

 

[Frank_And_Beans]

Hi OzzieGeorge,

I'm not all that savvy on this SV9100 so I appreciate your patience. As for a line key, in MB 15-07, I have assigned Function Keys 09-16 as "*01 - Trunk Key" = 1-8. I am assuming this is what you mean by a line key? I set my phone's buttons to each of the SIP trunks because when we first started having issues with the SIP trunks not being registered, I thought it might be because all the SIP trunks were in use. I found that wasn't the case but did not remove the programming as I like being able to see how many active calls we have. We have 8 SIP trunks BTW. When I press on of these keys, I instantly get a dial tone and am not required to dial "9" like I normally would to make an outside call.

So when I engage our SIP provider to switch from the PBX registering the SIP trunks to the SIP provider disabling registration and accepting our Static IP, we can receive calls, but cannot make them. I have not tried pressing a line key though when the registration is disabled, I have only tried making calls by dialing "9" and then the number. Unfortunately, I've been unable to get this to work so we revert to SIP registration to resume making calls.

For a little background, this all started because over the past 6-8 months, the SIP trunks randomly go down and it takes 30 minutes for them to re-register. It is odd that it is exactly 30 minutes too. If I reboot the PBX using "2nd initialization", the trunks re-register immediately upon reboot avoiding the 30 minute downtime.

The whole thing has had our PBX vendor stumped and they are working with NEC. I've submitted several backups of the SV9100 as well as dozens of wireshark captures. Due to being unable to resolve the issue, the PBX vendor suggested not using SIP registration and having the provider just accept traffic from our static IP but we can't even get that to work.

Sorry for the back story and long post. I really appreciate you and everyone who's contributed to this problem. Any further suggestions or questions are welcome! Have a great day!


Frank

 

[CoralTech]

If you are losing registration it's timing out. This is, I am almost 99% sure your firewall causing this issue. Firewalls will shut down streams that are inactive for X amount of secs/minutes. So both sides see everything is ok but the firewall has shut it down. SV9100 can be set to re-register more often than 30 minutes.

 

[Frank_And_Beans]

Hi CoralTech,

I've worked with our Ubiquiti vendor but I can open a new case just to be sure. The Firewall has been setup SIP ALG disabled and the UDP settings (UDP Other and UDP Stream) have been set to 300 seconds. The SV9100 has had the registration duration set to 3600 seconds (MB 10-29-15).

Are you aware of other settings that should be modified/verified? Appreciate your time and effort.

Thanks,


Frank

 

[Frank_And_Beans]

Hi Again,

One other thing of note. I still struggle to see this being a Ubiquiti problem because of how the issue plays out. As I mentioned, the SV9100 will lose registration for 30 minutes. I've attached a photo of the System Alarm page showing the disconnect and reconnect precisely 30 minutes later. If I restart the PBX using 2nd initialization, the SIP trunks reconnect as soon as the system comes back up. This is the only way I've found to get connectivity restored other than waiting the 30 minutes. My point being that if this was the Ubiquiti firewall closing/blocking the UDP connection, why would it take 30 minutes to re-establish when the UDP settings are set to 300 seconds (5 minutes)? Why would restarting the PBX circumvent the 30 minute delay?

I could try rebooting the Ubiquiti the next time the SIP trunks are down to see if a firewall reboot has the same effect as rebooting the PBX? Thoughts?

 

[CoralTech]

Simple, when you reset the system it re-registers. Another simple test is bypass the firewall with a simple router. Problem goes away it pretty much narrows it down.

 

[bk Patel]

hello all
just a thought
10:33:01 and 84:14:08 change this program value and check, may be come difference result !

 

[Frank_And_Beans]

Hi BK Patel,

In 10:33:01, the value is 3600 but in 84:14:08, the value is 1800. That is 30 minutes. I can try to test that but I need to do a bit of reading the exact purpose of this setting. Thank you for your assistance!!!

 

[OzzieGeorge]

Also check 10-29-15 and it may be worth checking 10-29-19 to keep the link alive!

 

[Frank_And_Beans]

Hi OzzieGeorge,

10-29-15 = 3600
10-29-19 = Unchecked

Should 10-29-19 be checked? Also, does BK Patel's suggestion about 84-14-08 relate to this issue or does that setting have a different purpose.

Thank you all!

 

[SomeGuyOnANet]

Hello, I'm experiencing an extremely similar issue as yours. I'm running an SV9100 as well as an Ubiquiti EdgeRouter. We also use Edge Communications. Our trunks have been disconnecting every 30 minutes or so out of nowhere. This system has been operational for nearly 4 years with little issues. This happened out of nowhere with no changes being made in our network. I'm at a complete loss as to how I am going to resolve this. I originally thought this was an ISP issue but I'm starting to question my logic now. We tried static Public IP and was able to bandage the issue for a couple of months with some very odd happenings such as having to dial out multiple times to get a tone to the phone not ringing at all but still connecting.

Has any solid resolution been found?

 

[Frank_And_Beans]

Hi SomeGuyOnANet,

I never was able to get my SIP trunks to use the Static IP registration. No matter what I've tried and worked with our vendors. The closest I got was I could receive incoming calls and Edge could see the trunks up. The problem was the PBX still was attempting to register the trunks and since it couldn't, I would get a busy tone whenever trying access the SIP trunks. It was like the PBX was waiting for registration before attempting to access the SIP trunks. That's the best way I can explain it.

The strangest part was after Nov 30th, the trunks have not gone down once. The problem magically went away and I still can't figure out why or how. I'm using a static NAT via JSON file in the Ubiquiti USG 4 Pro. We have AT&T fiber as our ISP. Edge as our SIP provider. Ubiquiti switches. Also in the Ubiquiti Cloud Key, I've disabled SIP ALG, set UDP timeouts to 300.

Wish I could be more help. Happy to share any SV9100 settings with you if you want to compare things.


Frank

 

[CoralTech]

Look to see if the edgerouter has been updated recently and make sure SIP alg is off.

 

[SomeGuyOnANet]

I will check SIP alg but the firmware is the latest revision 2.

Also I had the same thing happen. My sip trunks worked until a few days ago. Then they reverted back to the old failure to register.

Low Voltage Connoisseur

 

[SomeGuyOnANet]

Sip ALG was disabled within the edgerouter. I'm almost thinking its a timing issue with our SIP provider at this point

Low Voltage Connoisseur

 

[Frank_And_Beans]

Hello All. I'm revisiting this. We had several power outages over the weekend and some of our configuration was corrupted and the issue has started back up. I don't know if that is coincidental or directly related. We did also upgrade our ISP to a faster speed, different IP block but I'm not really sure how that could be a factor. Honestly, I would love to get away from the sip trunks needing to register but I've never been able to make this happen. Multiple failed attempts with the local NEC vendor and SIP provider. Something in the SV9100 just doesn't seem to allow this to happen. I am taking a shot in the dark by bumping this thread in hopes that someone has also had this issue and resolved it. Thank you in advance!

 

[ngstan]

We also have this issue. SV9100 with Edge Communications. Internet is 100MB up/down fiber. Router is a Cisco though. Same 30 minute issue. Sometimes it goes days without any issues and then like today, it's gone down 4 or 5 times overnight.

Instead of restarting the SV9100, I was instructed by our original vendor to go into 10-36, uncheck the box, log out, go back into 10-36, check the box and log out. Connects right back up after that. Still frustrating to deal with.

Did anyone ever figure out if the suggestions of 10:33:01 and 84:14:08 from "bk Patel" could impact this at all? That 1800 (30 minutes) in 84:14:08 seems tempting to change. If it's going to do this, it would be nice if I could have it come back in 5 minutes instead of 30.

 

[CoralTech]

Firewall is shutting it down. Up your timers there.