DHCP lease time

Hello Toradex,

I am using DHCP on ethernet adapter. I just set the registry key EnableDHCP to 1. All is working fine. Once the dynamic IP address is assigned by the server, the lease time is set to 1h 40min in my case. Should I take care about renewing IP address or all the stuff is managed automatically for me? Currently, after the lease time, it looks like the IP address is not renewed and the communication via ethernet becomes broken. Thus I would like to get information from you, whether it is supposed to take care about renewing IP address or not.

Thank you for help

Best Regards

Vlastimil

Dear @vpetrucha

The lease is renewed automatically.
There’s a known issue, that the Lease Expiration Time is displayed wrongly on all WEC7 platforms. We reported the problem back to Microsoft, but didn’t get a fix yet.
This is only a display problem, the actual lease renewal works as expected.

If you really experience communication errors after the lease time, this would be an unknown issue to us. In this case I would be interested to know:

  • Can we easily reproduce your network setup here ?
    (How does your network look like)
  • Do you see the same failures if you load CE6 onto the Apalis T30?

Regards, Andy

Hello Andy,

Finally I found the root cause. We had accidentally connected also a router to our network. This caused there were 2 DHCP servers available. As a response to the lease time prolongation request (issued by ApalisT30) the first answer was received from the DHCP server which didn’t assign the IP originally (so it returned DHCP NAK). The next packet was from the 2nd DHCP server (it returned DHCP ACK as expected). It looks like ApalisT30 accepts the 1st answer only and consequently behaves like the lease time was not prolonged. This causes the connection is broken once the lease time is elapsed.

So it should be easy to replicate such behaviour at your side. I don’t know whether it is a really issue or not, I would say yes.

Best regards

Vlastimil

Hello Andy,

In my case the next request has been issued in 37 minutes from the previous one, but with the same scenario. Actually we have to re-connect connection each 1h 40min which is very unconfortable in our product.

Now we know the issue, but our product is world wide so it would be benefical to fix the issue to preceed a customer’s troubles. Unfortunatelly I cannot judge whether it is common to use more DHCP servers in network. I would expect it is rarely rather.

Best regards

Vlastimil

I forgot to mention the next workaround is to use static IP.

Regards

Vlastimil

Hi @vpetrucha

It seems the problem is a combination of various issues:

According to information we got from CISCO, multiple DHCP servers should communicate to each other, and only one of them should reply to the renewal request.

The fact that the Apalis T30 ignores the ACK after a NAK is also correct - however, after the received NAK, it should start a new DHCP request from scratch. If I got you right, this does not happen.

Additional Information: The DHCP behavior is explained in RFC2131.

Is this an issue for you which needs to be investigated, or did you solve it by disconnecting the 2nd DHCP server?

Regards, Andy

Dear @vpetrucha

We decided that this is an uncommon DHCP server configuration, hence we will invest our resources into more valuable features and fixes. This was the first report for such a DHCP issue that we ever got, and it was in a accidentially wrong configured network.

Of course if the problem becomes important in the field, or if more users report this to be a problem, we will shift our priorities.

Regards, Andy