All extensions unregister and re-register simultaneously

[JohnS_3CX]

Hi @Jagk23

The correct ports are here, provided you did not change them from the defaults during setup.
https://www.3cx.com/docs/ports/

If the firewall checker fails then you need to address this as your number 1 priority.
In case you do not know how to open ports on AWS, then you can take a full backup of your 3CX system and then deploy it it via PBX Express which takes care of everything for you.

Here are the instructions to do it via PBX Express:
https://www.3cx.com/docs/cloud-pbx-amazon-aws/

-OR-

Here are the instructions to do it manually via AWS Marketplace:
https://www.3cx.com/docs/pbx-amazon-aws-marketplace/

Once you are up and running and your firewall checker passes, then we can look at your phone provisioning

 

[Jagk23]

Hi,

I've corrected the port issue and my firewall test now passes as expected.

Regards,

Jack

 

[JohnS_3CX]

Great news Jack! Now lets focus on provisioning.

Since this is a cloud PBX there are only 2 provisioning types that you can use:

1. STUN Provisioning
2. SBC Provisioning

NOTE: "Local" is out of the question unless you have a VPN between the phone's site and the PBX, so you should not really have any phones on "Local mode" right now.

a)First make a list of how many sites you have and how many devices per site.
b)Then make note of whether you can control the firewall at each site or not.
c) STUN is ok if you can control the firewall at that site and IF you have 2-3 phones maximum. If you cannot control the firewall, then you will need an SBC for reliability
d) For sites with more than 2-3 phones, SBC is a better idea (whether you control the firewall or not) since managing a lot of ports can be a pain.

 

[Jagk23]

Hello,

We have a VPN as previously mentioned between the site with Local phones and the PBX.
We do not have more than 2 - 3 STUN phones per site. Most have a single STUN handset - there is one with 3 handsets.
All firewalls are under my control and are configured correctly.
It doesn't matter whether the phones are connected via STUN or Locallly - the extensions still drop out.

Thanks,

Jack

 

[AWS2P]

how did you setup your stun's phone , have you set a specific sip port to each phone? phones have static IP, RTP range unique for each phone?
Settings done accordingly on local site where phones are used?

 

[JohnS_3CX]

Hi Jack,

- What are the specifications of the AWS machine (cores / ram / storage ) and how much free space does the management console report currently?

- What is your license type/size?

- Do you notice high CPU usage or get emails about services stopping?
(Under Settings>Emails>Notifications>3CX Services are stopped or fail)

- Also how many users connect over webclient /desktop client / mobile client?

 

[Jagk23]

Hi guys,

1. Each STUN phone has a unique set of RTP ports and IPs individual to each site, the problem isn't isolated to the phones configured using STUN.
2. The AWS instance uses an m5ad.large instance which has 8GB RAM and 2 cores, the disk space is reporting 32% used and 20GB free on the management console. Recordings etc are on a separate drive with around 50% free.
3. We are using a Professional license
4. No there is not any high CPU usage, no services stop.
5. We probably have around 15 users who use the webclient, nobody uses the mobile or desktop client.

Jack

 

[JohnS_3CX]

If everyone simultaneously re-registers (from all sites) then we can't really point the finger at a specific site for now.

It sounds as if the PBX stops responding or becomes unreachable to everyone every once in a while if they all appear to re-register at the same time.

I would be looking at a m5ad.xlarge if your total extensions are above 10 and up to 50 in total.
https://www.3cx.com/docs/recommended-hardware-specifications-for-3cx/

Also since you took care of the firewall issue, monitor the PBX a bit further and see whether this affects the problem you were facing. You can additionally enable Verbose logs on 3CX to see if we can catch any events that tell us what's happening

 

[Jagk23]

It's 100% not a specific site or connection type issue.

I can increase the specification at a later date but it doesn't seem likea resource issue to me.

The issue has occured since fixing the firewall issue. How can I enable verbose logging?

 

[JohnS_3CX]

It's under Activity Log > Settings > Verbose and then press OK.

You should restart services after doing so. The next time it happens, start looking through logs to see if anything stands out.

One thing that can mess with registrations when the network conditions are healthy, is time drift on either end (phones site or PBX machine). This can vary from model to model or site to site and even though we provision phones to use a common NTP server, in case it is blocked or in case your local DHCP server sends them some other time, then eventually the registration expiry won't match. Same goes for the PBX machine, if the time drifts or is unstable, the same scenario will happen and eventually registrations will not sync. This can also happen when you have some other service or monitoring tool installed on your PBX and it may sync the time every now and then with a different NTP or offset. Sometimes this is even visible on the 3CX dashboard CPU graph if this too bad