Test LAN to troubleshoot some problems

[Caple]

Hello everyone.

I'm working for some months on a new job, where we use a helpdesk system, similar to Zendesk, to communicate with costumers. We changed from regular phone lines to voIP, using the same system, about 45 days ago.
We always had some small problems, like emails not fully loading and options not loading when working on a ticket. From time to time some calls would be dropped and/or the costumer couldn't hear us.
Other offices, in different locations globaly, use it with no problems.
Working with the support team from the helpdesk company to troubleshoot it, they were able replicate one of our problems when limiting the web browser to utilize less internet speed. They said the problem is on our side and is related to poor internet quality.
We have a 50Mbps dedicate fiber link and, before reaching the ISP to complain about it, I'd like to grab a few tips on how to test my LAN. I'm asking that because we have a pfSense box, with a not up to date firmware, that was installed some years ago and never maintained.
So, I'd like to test the LAN before contacting the ISP. Specially because we already contacted them regarding similar problems and they said everything was fine on their end.

Right now we have a pfSense, running version 2.1. All computers (around 60) are connected to four 10/100Mbps switches. We have only one LAN port um the pfSense box, so the internet is routed from switch to switch.

No wireless connections and no other hardware connected to the LAN. Only PCs and printers.

Any input on this would be really helpful.

Thanx.

 

[Caple]

Some more info.

I ran iperf on some machines here. The result was always above 70mbps.
A tracert to the helpdesk location, in the UK (we are in Brazil) does not show anything strange. 158ms to it.
Pinging it for about 1min resulted in 0% packet loss.

 

[sfx2000]

Well, first open a trouble ticket

It might be the pfSense box - 2.1 isn't that old, but it's getting there - current release is 2.3.3_p1 (as of a couple of days ago).

What hardware is pfSense running on?

 

[Caple]

It's a Pentium G2020. It's a dual core CPU, with 2GB RAM. Never seen it using more than 12% of its RAM and a couple of % on the CPU.
All it has configured is DHCP server and a couple of static IPs.
The only difference from the original install is that we changed our ISP (prior to using voIP) and the WAN settings were changed. But I assume those leftovers would not risk anything. See attached file.

 

[System Error Message]

VOIP is vulnerable to packet drops. So if your side uses digital rather than phones from rj11 than that helps. First you should check your upload, is it 50Mb/s? Check your VOIP software quality, the default is 56kb/s, you should change it to as high as possible but no higher than what the phone companies use as that means data gets lost. You should adjust the bitrate to highest the minimum phone companies use. Other ways to set the quality is with a high bitrate but lower audio quality to give it more space, similar to live video encoding where lower resolutiony and higher bitrate lose less information and is smoother.

If all the PCs are wired up than its unlikely to be a matter of LAN. make sure you've set your filters and QoS. If you are using dual WAN that make sure to use PCC load balancing. Perhaps the router still considers the left over settings of the last WAN to be in use and makes it a valid gateway.

 

[Caple]

VOIP is vulnerable to packet drops. So if your side uses digital rather than phones from rj11 than that helps. First you should check your upload, is it 50Mb/s? Check your VOIP software quality, the default is 56kb/s, you should change it to as high as possible but no higher than what the phone companies use as that means data gets lost. You should adjust the bitrate to highest the minimum phone companies use. Other ways to set the quality is with a high bitrate but lower audio quality to give it more space, similar to live video encoding where lower resolutiony and higher bitrate lose less information and is smoother.

If all the PCs are wired up than its unlikely to be a matter of LAN. make sure you've set your filters and QoS. If you are using dual WAN that make sure to use PCC load balancing. Perhaps the router still considers the left over settings of the last WAN to be in use and makes it a valid gateway.

The audio quality is not a problem. Some dropped calls and inhability to answer some calls are. Along the errors when loading tickets.
The calls are answered on the webpage, we are not using RJ11 phones. Just a headset.
I'll try to prioritize the access to the website, by IP, on pfSense Traffic Shaping options. Dunno if it has that option.
Maybe the problems will, at least, diminish.

 

[System Error Message]

in this case i would also check the website. If your upload is fine and you've done QoS than check the website and the website's provider. pfsense has QoS for sure.

 

[Caple]

in this case i would also check the website. If your upload is fine and you've done QoS than check the website and the website's provider. pfsense has QoS for sure.

I'm looking at QoS on it. I'm not sure if I'll be able to prioritize the website's IP on it. But prioritize voIP is a possibility.

 

[sfx2000]

If you can - pfSense does have good traffic shaping/quality of service - check the wizard under "Firewall" -> "Traffic Shaper" - use the use the Traffic_Shaper_Multi_All.xml

 

[sfx2000]

I would recommend updating to the current 2.3.3 release - not just functionality, but also many security fixes...

Make very sure to back up the config first, as that jump can be troublesome from 2.1 to 2.3...

Good info over on the pfSense community forums..

 

[Caple]

Hello.
The changes I did on the router didn't help. I got one of the developers to connect here on Team Viewer and he took a look.
Seems like the loading for a file, which is small - only 2,3mb, is timing out. He will try to route us directly to another server. Right now our connection to the helpdesk system is going to a Texas server and, from there, to UK.
In the end, we don't really know where is the problem and why it's timing out. But let's see if that change of route will help us.
Thanx everyone for the replies.

 

[Caple]

Well, routing us to another server didn't make a difference.
All the small files are loaded properly, but some bigger ones (still small, around 2,5mb) are timing out. Sometimes, after two or three tries, it loads under 30 seconds and everything is ok.
What should I test next? iperf tests seemed just fine. Maybe the 10/100 switches are being the bottleneck? We have three and they are all full. No ports left.
I couldn't upgrade pfSense, yet.

 

[Samir]

Before you go spending a lot of time on the pfsense install, since your routing needs are quite basic, get a simple router that can do what you currently have--dhcp, static ips, and wan to lan >50Mbps. Configure it identical to the pfsense box and then swap it into place. See if that changes anything.

If the problems go away, the issue is related to the pfsense box. If the problems persist, it's definitely not the pfsense box and you don't have to spend any time on it with an update to resolve this particular issue.

 

[Caple]

Before you go spending a lot of time on the pfsense install, since your routing needs are quite basic, get a simple router that can do what you currently have--dhcp, static ips, and wan to lan >50Mbps. Configure it identical to the pfsense box and then swap it into place. See if that changes anything.

If the problems go away, the issue is related to the pfsense box. If the problems persist, it's definitely not the pfsense box and you don't have to spend any time on it with an update to resolve this particular issue.

Good ideia.
I'll try to get my hands on a decent router to try. Nothing fancy.
Thanx for the suggestion.

 

[Samir]

Good ideia.
I'll try to get my hands on a decent router to try. Nothing fancy.
Thanx for the suggestion.

Yep, nothing fancy. Let us know what happens.

 

[Caple]

Yep, nothing fancy. Let us know what happens.

The only thing I have right now is a MikroTik RB951Ui-2HnD. I have to check how to configure that properly.

 

[Samir]

The only thing I have right now is a MikroTik RB951Ui-2HnD. I have to check how to configure that properly.

That should work fine, and user System Error Message on here will probably be able to assist you if you run into any trouble as he's the local Mikrotik guru.

 

[coxhaus]

Are the switches 10 Mbps on the ports with 100 Mbps uplink or all the ports 10/100? How are the switches connected? Are you using some kind of up link port so a workstation does not have the same priority as a whole switch? Are there certain PCs having more problems than others?

It could be time for a switch upgrade. If the old switches are the bottle neck and you upgrade, it is possible you will move the bottle neck to the router. But it would be better to have the bottle neck at the router than in the local switches. GIG switches are fairly cheap now. If you have uplink ports on the current switches maybe you could try a GIG switch coming off the router and plug the 10/100 switches into it making sure no work stations or slow printers are plugged into the GIG switch.

You may find you have more than one problem. You are going to need to try things. A modern PC can move a lot of data now days when you are talking 10/100 ports. VOIP packets are UDP streams without any error correction so bottle necks can cause problems with UDP streams.