Finally, cams are installed

[IAmATeaf]

OK something pretty strange going on which I don't really understand. When I continuously ping the ip address of the network card that sits on the side of the cams no issues, all pings come back in under 1ms, left it going for a 10-20 seconds. If I do the same for any of my cams then I intermittently get a "Request timed out" message but all the successful pings are less than 1ms.

I've checked the config of all the cams, I have the default gateway set to something that doesn't exist ie 192.168.1.1, I've updated the network adapter drivers as well.

Not too sure what to do next, how do I find out if it's the POE switch of if it's the cams? The fact that they all do it sort of points to the POE switch but when pinging the network card which is connected to the same switch it's all OK.

Anybody a networking expert here that can steer me towards what I can try next. I'm thinking I might connect another device (such as another desktop) to the POE switch to see if that behaves the same?

 

[Sphinxicus]

since all your cameras are suffering disconnects i guess it makes sense to start at the PC's interface and move from there.

When you pinged your IP address of your network card from (I'm assuming the same PC here), no frames make it onto the wire so you are just testing the TCP/IP stack out.

Open a command prompt and type netstat -e -t 30

This will give you details about your interfaces and will refresh every 30 seconds. Keep an eye on Discards and Errors. If they are increasing then the first thing i would do is replace the cable between your PC and switch with a known (pre-made) working one, not one that you have crimped yourself. Also check the workload of the PC. If its overworked it could start dropping frames or reducing its window size to tell sending devices to back off.

If none of that helps or shows issues then you could run wireshark on the PC to capture traffic between the PC and the cameras. Looking at the capture you will get a better idea of what is going on. Weather it be your PC dropping its TCP window size to 0 or multiple TCP re-transmissions.

To see if there is an issue with throughput with the switch or workload with your PC, while running a wireshark capture you could also try unplugging all of the cameras but one and see if the same issue occurs. If this camera remains stable then add another and keep going until it starts happening. When it starts happening take a look at the capture to see what happened at that time.

Just throwing that out there, someone else will probably come along with a less messy response

 

[IAmATeaf]

@Sphinxicus

Much appreciated, I thought I had found the problem, the switch is a managed switch and I hadn't set its IP address but that didn't resolve it. I also just upgraded the firmware of the switch but again same issue. One thing I have noticed is that even accessing the Web GUI for the switch is really slow, so I pinged it and it has the same timeouts.

Just ran the netstat command and Discards and Errors steadily grow, both numbers perfectly in sync with each other.

The switch is in the loft so when I'm next up there I'll replace the patch cable, current cable is a pre-made 0.5m patch cable so I'm going to replace it with a pre-made 1m cable to see if that improves things.

 

[adamg]

You might be hitting bandwidth limits of your network links. Have you noted the utilization/bitrate on all links?

 

[IAmATeaf]

Phew sorted.

Climbed into the loft with replacement cables, my Poe switch has 2 uplink ports so decided to plug the new cable into the 2nd lan port of the dual network adapter on the pc. That’s when I noticed that the existing port had an amber light indicating that it was connected at 100mb. Strange as the desktop thought it was connected at gigabit

So i unplugged the existing cable and now use the spare network and uplink ports then disabled the original port on the desktop from within windows.

Did some ping tests and net stats and now no errors and I can now actually logon to the Poe switch whereas before it was completely unreliable with all the dropped packets.

 

[IAmATeaf]

Damn what an improvement. Before I was getting very stuttery video and some of the cams would drop to a few frames a second, now all are steady at 15fps.

Now need to work out zones as at the moment set to basic motion I’m getting hundreds of alerts a day.

 

[Sphinxicus]

Phew sorted

Glad you got it sorted! Funny when two endpoints claim to be happy with speed/duplex negotiated but in reality are blatantly not in agreement. I have a switch-port on my Cisco 2960G that will not for the life of it negotiate 1000/full with any other device. Ive just put it down to being a duff port now.

 

[IAmATeaf]

Yes it was pretty strange. The desktop thought the net port was running at gigabit and so did the web gui on the switch but the light on the switch was different. Might investigate further when I have more time but just switching ports on the switch and desktop resolved it so that’s good enough for the time being.

Checking this morning everything is still working fine.

 

[Sphinxicus]

to be fair, it could have negotiated a gigabit connection, the amber light could be indicating that the port is in an error/alarm state. With it being a managed switch it will record similar metrics to the NIC on your PC so will know if its dropping frames and will give you an amber light to attract your attention. The manual for your switch will explain what the colour means, the GUI may also have alerts for that port available. I guess it sounds like a bad switchport/cable/connection.

Either way its working now so "mustn't grumble"