Erratic behaviour after 2-3 weeks of runtime

Status
Not open for further replies.

etsiot

Member
Jan 14, 2018
46
1
8
54
Hello All,

We have been running a FusionPBX install for some time now.
Some erratic behaviour is observed if the server is not rebooted every 2-3 weeks.
Some of these problems are:
- Doorphone may not ring at all
- Attended call transfer may not complete; i.e. secretary hits transfer button but the call is not transferred
- Call audio may experience initial delay; i.e. you answer the call but may have to wait 3-5 sec before you can hear each other
- Second incoming call not appearing in LCD screen

Problems existed in 4.2; upgraded to 4.4 but they still pop up.
Have tried to check the logs but nothing apparent shows.
Overall CPU load is minimal, so can't be the reason. Memory neither.

I have decided to cron a "service freeswitch restart" every morning to check if the issues go away.
Yet, it still puzzles me why this happens
Any clues would be appreciated, because it's really frustrating for the users.

Thanks in advance,

Vangelis
 
Last edited:

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
I do not believe any of these issues will be related to FusionPBX. I have boxes doing thousand s of call per day that have not been rebooted in well over a year nor Freeswitch even restarted. I wouldn't mind betting they are all NAT related in some way.
 

etsiot

Member
Jan 14, 2018
46
1
8
54
Thanks for the feedback.
I would expect both FusionPBX and fs not to require a restart, hence my question.

I am running with nonat, and all traffic is internal, can nat still cause problems in such a scenario?
Could openssl or related ssl libraries create issues? At some point I used that box to run openssl, are known issues with it?

Many Thanks,
 

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
So the box is on an internal LAN with dual interfaces and a public IP on one of the interfaces??
 

etsiot

Member
Jan 14, 2018
46
1
8
54
No, PBX and all phones (Yealink), plus doorphone (Fanvil) run in private subnet (192.168.x.x.)
 

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
So what happened to: "I am running with nonat, and all traffic is internal, can nat still cause problems in such a scenario?"
 

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
If your IPs are private and your PBX talks to the internet then NAT is very much involved.
 

etsiot

Member
Jan 14, 2018
46
1
8
54
Let me give a topology, sorry for not including at first place. No SIP traffic with internet. All calls are through SIP gateway. Phones are all Yealinks
Schematic-2018-12-18.png
 

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
Next time you have the problem, maybe just try restarting your gateways
 

etsiot

Member
Jan 14, 2018
46
1
8
54
I am not using any SIP carrier, calls are PSTN only terminating at the Grandstream GXW4108.
Also, issues go away when restarting the FusionPBX box, don't have to restart the gateway.
 

DigitalDaz

Administrator
Staff member
Sep 29, 2016
3,038
556
113
Oh, I see, pure LAN, then rather than restart it, use sngrep to see what is happening when you try and do whatever it is. I still don't see it being a freeswitch issue.
 

bcmike

Active Member
Jun 7, 2018
326
54
28
53
This could be absolutely anything, you really need analyze both the network and your hardware as the problem is happening.

You could just be running into network congestion on your LAN caused by something that happens at the same time once every 2 weeks or once a month. A backup, windows update, anything like that. Check all your scheduled tasks and Cron jobs. It could be as simple as big file transfers to a file server, etc..If you can, implement QOS on your switch and if its a gigabit switch turn on flow control.

You may also be having some sort of hardware issue that's not entirely apparent, again you have to catch it as its happening and look at your CPU and memory utilization. Also bad memory causes a lot of sysadmins to chase their tail. If you can run memory checks after hours. One last thing make sure the disks aren't crapping out, your PBX might be waiting to write a cdr or voicemail and causing things to go sideways.

Good luck, intermittent failures are the worst to troubleshoot.
 
  • Like
Reactions: etsiot
Status
Not open for further replies.