I am using a RB750 (Router OS 4.5) in one of our offices, which has some linux based machines provided by a 3rd party company attached to the router, using DHCP. Now after 2 or 3 weeks since this office is up an running, I am hearing of reports that the linux machines "lose their access to the internet", and the only way to resolve that is by releasing and renewing the lease (manually).

The company that provides the linux bases machines says, they experienced similar problems when using other "low budget" routers. They suspect that there is a problem with the DHCP server losing track of which IP has been assigned to other machines, causing an IP conflict - especially if all machines are turned on simultaneously via a central power-switch, and the several leases are obtained within a short time-span.

Is it possible that such a problem exists on the RB750? I am very happy with the device, and although it is low budget, my impression of it is very solid and professional. Searching the forum did not reveal any reports of other users with a similar problem, and I would have thought that a function such as DHCP should be rock solid.

Of course it is possible that there is a configuration problem, although I leave the default dhcp-configuration untouched, only add an IP-pool and configure the dhcp-server network - but both settings are trivial and benign. I did find the "authoritative" setting, which is currently at 2 sec. I figure I could set it to "yes", since the RB is the only dhcp server inside the network. Also I wonder if it makes sense to tick "add ARP for leases", although ARP entries are already added dynamically.

Any tips how I can test if there actually IS an IP conflict, and what might cause this?

This could also simply be caused by someone else having configured a static IP from the pool after it had already been dynamically assigned. Do you have switches that allow you to look at their MAC address tables? When there is a duplicate IP check everyone's ARP table and see if a different MAC address maps to it, and track it down on the network.

I'm having a similar problem. I have a RB411A AP doing dhcp to customers on that access point. Lately, we found that customers (mostly routers) were getting ip's, but they were declining them as they thought they were in use. Oddly, I found one customer who's Last IP on the Registration table kept changing to everybody else's IP's. Thinking it was a bad router, I disabled the ether1 port on his radio (RB411 with XR3) and the problem went away. I then went out and replaced the guys router with a new one and all seemed fine till today. Problem happened again, I disabled the ether1 port on the customers CPE and it seems fine again.

AP is at ROS 4.7, I just upgraded the customer CPE to 4.7 from 4.6. So far, luckily, this is only happening at one AP. I'm reluctant to turn on the ether1 of the customer for fear this problem will happen again, but I'm afraid I'll have to. The AP has been running for quite some time without any issues like this one. I just started happening for some reason.

Tom

I'm having a similar problem. I have a RB411A AP doing dhcp to customers on that access point. Lately, we found that customers (mostly routers) were getting ip's, but they were declining them as they thought they were in use. Oddly, I found one customer who's Last IP on the Registration table kept changing to everybody else's IP's. Thinking it was a bad router, I disabled the ether1 port on his radio (RB411 with XR3) and the problem went away. I then went out and replaced the guys router with a new one and all seemed fine till today. Problem happened again, I disabled the ether1 port on the customers CPE and it seems fine again.

AP is at ROS 4.7, I just upgraded the customer CPE to 4.7 from 4.6. So far, luckily, this is only happening at one AP. I'm reluctant to turn on the ether1 of the customer for fear this problem will happen again, but I'm afraid I'll have to. The AP has been running for quite some time without any issues like this one. I just started happening for some reason.

Tom

Sounds like proxyARP is enabled on ehter1.

I ran across this earlier when searching for clues. It is not on the client or the AP. I'll go into every customer on this AP and check them out. Thanks for the reminder.

Tom

Coming back to my original problem with "possible IP conflicts", or what ever else reason is causing the linux machines losing their connection to the internet - I discovered the following behavior: after rebooting these machines (and obtaining a fresh DHCP lease), they work for a couple of hours. But then I notice that I can't ping some of the machines anymore (from the router). Probably at that point these linux machines are "dead" too, but I can't confirm that since I am only connected remotely to the router. Some of them are (presumably) still alive, cause I can still ping them. There's also the one or other none linux machine (like a windows machine and a network printer), which keep working all the time.

The lease is issued for 3 days, so I don't think that this particular problem is caused in any way by the DHCP mechanism, since it happens a few hours after the lease was issued.

The other peculiar thing I noticed was that the ARP table is shrinking pretty quickly - entries get removed automatically. Even for machines that are still working fine (like the PC). After pinging one of these alive machines an ARP entry is added again, but disappears shortly (approx 10 - 60 mins) after. I did not notice a "shrinking ARP table" on any of the other routerboards that I have control over.

Thought this might be a hint that points towards the actual problem?

It was not my intention to hijack this thread, but to offer a "similar" type of issue with DHCP and ip conflicts and dhcp denials.

My apologies.

Tom