Some users/routers often can't access the internet [closed]

Question

There are times that some users can't access the internet. Usually this happens with the routers with DHCP enabled, they alternately can't access internet. I suspect that the main router is faulty. I need your opinions. Here is the network setup.

NOTE:

PCs/laptops obtain their IP addresses automatically

Main router: 192.168.10.1

Other routers: 192.168.11.1 to 192.168.14.1

Their channels are different based on location. 1 , 6 , 11.

All switches are unmanaged.

Boxes represents different offices.

Answer 1

You need to do some debugging to troubleshoot this problem.

Problem like intermittent access to the internet or other networks in your design are sometimes caused by:

1. The router is simply overloaded
2. Because you said you use DHCP, there could be IP conflict issues if some users use static IP
3. The 192.168.10.0/24 network (Main router) seems to be quite large that you even cascade some 24-port switches together. Remember that broadcast will be very often and so broadcast storm could happen often.
4. Wireless interference, but this is unlikely because you said you use different channels
5. Your devices could be faulty, but this is also unlikely

This way it is better to do some actions:

1. Monitor the routing tables. If all routers are consumer grade routers you usually do not have access to these tables. If you happen to have access to those, monitor those routing tables. See if there are any dynamic changes happening from time to time. Dynamic routing protocols are the main cause of this. Conflicting routing protocol settings can confuse the routers in a network and periodically change the routing tables. But if you use only consumer grade routers, those routers are sometimes not equipped with configurable dynamic routing protocols. Instead, all routes are fetched through DHCP, giving them a default route. If this is the case then you can skip this step as you won't have access to the routing tables in consumer grade routers aynway, unless you use other router firmware.

2. Make sure the DHCP settings are correct. Your network seems to use /24 mask. This means 254 IPs can exist in a network, including the gateway. This does not necessarily mean that the DHCP provides enough IPs for the users. The DHCP pools could be full. If the DHCP pool does not provide enough IPs for the users, users simply do not have IP to connect. You can simply enlarge the IP range of the DHCP. If there are some users that do not have connection to the internet since the first time they connect to the network, do check the DHCP settings.

3. To see if the internet connection is down I guess you do some ping test to some sites in the internet. Does the ping fails with request time out (RTO) or destination unreachable? If it's a request time out, packets sent from a user do not get any reply. In the case of conflicting IPs, packets can be sent to different locations randomly. Because there are two same IPs, routers can be confused to where packets are sent (the ping reply message is sent to wrong location). If it's destination unreachable, then it's a network problem. destination unreachable is generated by a router as a reply message to indicate that the destination is unknown to that router. The router does not know where to send it. Use traceroute to see where your pings are lost. Intermittent destination unreachable messages could mean that your routers are faulty. It could mean that the router cannot parse the routing table completely (for example full memory), or other problems that cause the destination address to intermittently get erased/written from/to the routing table.

4. See if the users that do not have access to the internet can ping: their default gateway, the main gateway, other users in the same subnet, or other users in different subnets. This is so you know which interfaces/routers are the source of the problem. Point is, you need to isolate the source of the problem by making ping to and from different locations from time to time. Do a ping for at least 5 minutes and imagine where the packets will go, while also imagining where the reply will go, sometimes your packets are sent to the destination but the reply does not get to you.

5. See if your ping packets reach the destination by using monitoring tools like wireshark or tcpdump.

6. Try to disable some subnet for a while and see if the problem persists. If it's not preferable to disable a subnet to reduce load, try to use iperf to load test your infrastructure. Place the server and the client in different locations across the subnets to see which routers are overloaded. If you see speed rate spikes the subnet or a network device could be overloaded.

7. Consumer grade network devices sometimes have small memory. Routing/forwarding tables are stored in the memory. Packets are queued in the memory. If the queue is not enough to fit the packets, some packets could be dropped. This could sometimes result in request time out. Increase the load in the network to see if you see more request time out. See if the number of RTO messages are linear with the network load.

8. A router such as D-link DIR-655 is probably too small for that network especially when many users are active. An overloaded router does not always make the users cannot connect to the internet. It could make many packets get dropped (like point 7) or queued for too long in the router. If this is the case you should see an increase in response time when browsing the internet when more users are active because many IP packets get dropped or queued for too long, and so many HTTP packets (TCP segments) get resent.

9. Consider using better main router for your network. You can replace the main router and switch 5 together into a layer 3 switch.

10. Many custom router firmware allows you to see the usage of your router memory. Consider using this if your routers firmware does not allow you to see memory usage.