What to do if some computers got the wrong IP address?

Good day!
There is a small office. There are a couple of dozen computers (mix of PC and Mac), connected in one local network.
There are two router from the same manufacturer. The approximate scheme echobrain in the diagram below:
A simplified scheme of the network
5afe8cac98124770563118.png

The distribution of IP is through DHCP.
80% of the cars working as it should. But some of them sometimes lose the Internet. Spontaneously.
When you go into the connection settings, instead of the standard subnet 192.168.31.* for some reason, have a kind of 169.*.*.*
This may happen with a Mac and a PC.
On Windows usually help the team
ipconfig /release
ipconfig /renew

After entering these commands on vindovsky machine, with some probability all can suddenly earn on all machines at once (this was exactly one time, maybe this is not the reason and just coincidence). Or just for her.
Assigning a static IP address and IP router in marshrutizatora in the connection settings of the machine is also sometimes either works or not.
Rebooting the router also either works or does not work, from situation to situation.
I have suspected the router, but I would like to make sure that there are no other possible situations and problems, price than to change it. For example, I heard that the car can begin to broadcast to the network that it is the most important here and it is starting to batter others. Like it or not - I do not know, do not have sufficient experience in this.
6 answers
June 5th 19 at 21:14
Solution
What aprutina problems with the network it's 100%. In settings instead of automatic address to register your on the tab "Alternate configuration", but the lack of network will not solve it.
June 5th 19 at 21:16
Solution
This week faced with the same problem. Peer-to-peer network on a dumb switch, router - everything. By process of elimination, found a bad switch, replaced - works.
June 5th 19 at 21:18
Solution
Two router handing out the same subnet via DHCP? Leave one router with DHCP, the second connect in bridge mode.
June 5th 19 at 21:20
Solution
Address 169.*.*.* Windows assigns itself when it cannot get an address from the DHCP server. As part of the computers with addresses everything is fine, problems are likely on the client side:
1. Problems with the DHCP client. Helps the command netsh winsock reset
2. The client requests an address, but doesn't get it. The firewall can cut the replies from the server. Add an explicit rule to pass incoming udp packets on port 68.

And, check to see what time displays the address, usually a router issue for an hour, or less. If a network of small computers, and small, can increase the lease time of addresses on the router, say for a couple of weeks. Then the clients after receiving the addresses will be to hammer the dhcp server for its queries.
June 5th 19 at 21:22
Solution
And why the second router?

A 169.* the network is an attempt Windows to work without the presence of DHCP (as above someone wrote)
The offices are at some distance from each other and some work with a laptop as often have to move from the office to the pavilion, for example - gustave_Ha commented on June 5th 19 at 21:25
Perhaps you have confusion with the purpose of the devices, and maybe I misunderstood something. Second router for WiFi?

If my guess is right and you stick a second router in the public network wan, that is your problem.
Output:
1. Not the right failover - disable the second router DHCP (if enabled)
2. The correct failover - configure the second router DHCP relay on the first (realistic for Mikrotik, openWRT and other smart devices)
3. Spike failover - configure non-overlapping ranges of addresses from the same subnet (might be a problem with the constant portapros address)

The UpLink between the two routers must go through the normal network ports.

If it is on the knee. - Joel commented on June 5th 19 at 21:28
the router are a distributor of Wi-Fi and a bridge of sorts - it is directly connected to the main switch, which at the same time connected 10-inch cable to the main router.
The second router is already podluzany remaining switches in the second study. In the first to the first, which is next to the router.
Try the options, but it will not be soon. In any case, the network will have to deal with, as the number of jobs becomes more and more.
A 169.* the network is an attempt Windows to work without the presence of DHCP (as above someone wrote)

On Macs the same nonsense happens sometimes - gustave_Ha commented on June 5th 19 at 21:31
draw me in more detail the network map (up to the ports on the routers), and models of routers. The option that you do not pass the DHCP packets from the I dismiss immediately, you have the same two points via VPN going? - Joel commented on June 5th 19 at 21:34
June 5th 19 at 21:24
These machines just simply do not find a DHCP server upon the next attempt of the updates addresses.

Try to understand why this is so, how much time waiting for a response DHCP server client machine, where the traffic can go the wrong way, etc.

Router 2 is just a bridge?

Find more questions by tags DHCPNetwork administrationNetwork equipmentComputer networks