So the DNS servers are NEVER wrong when manually renewing the lease?
If not, do the DNS server settings seem to change at about the time the lease renewal duration trigger? If so, I'd try to get a wireshark trace of that connection at about that time... If it's a random time.. then a trace with a trigger defined for DHCP conversations. --sc From: [email protected] [mailto:[email protected]] On Behalf Of Melvin Backus Sent: Thursday, April 24, 2014 9:08 AM To: [email protected] Subject: RE: [NTSysADM] DNS server settings getting changed No other DHCP servers that I'm aware of but certainly worth a look. That said, the process I'm using to detect the change reports the DHCP server (I'm just doing psexec and ipconfig) and they are all pointing to the correct one. The only difference we've found is that the DNS servers are wrong. We've even connected to those machines and manually checked settings to confirm they are still set for DHCP, etc., when it happens. The machines don't have to have publicly available IPs, only routable IPs. As in no NAT, and no private IP ranges. So, we've got IP blocks that we assign to all those machines. They never see the outside world, but they are routable to the outside should that need every arise. Think large, formerly monopolistic telco. J Essentially the same as above. In order to be allowed to see their DNS servers, they have to be on non-private subnets. Since we already have everything setup on private subnets for server and infrastructure, rather than move that, we added a couple of DNS servers on one of the non-private subnets. They just have a list of conditional forwarders and either forward DNS request to the partner, our domain DNS, or the Internet. I'm not sure that's the best way to handle the whole thing but it provided a mechanism for us to allow the systems which require access to the partner network to resolve it (we used to actually have to maintain hosts files because they didn't use DNS) without a total rework of our DNS infrastructure. Things would probably be different if we were starting from scratch, but that's almost never the case. J -- There are 10 kinds of people in the world... those who understand binary and those who don't. From: [email protected] [mailto:[email protected]] On Behalf Of Andrew S. Baker Sent: Thursday, April 24, 2014 8:52 AM To: ntsysadm Subject: Re: [NTSysADM] DNS server settings getting changed How sure are you that there isn't another DHCP server in the mix? Have you ever looked at the what DHCP server a machine with bad DNS settings has? Also, I must say that I've never seen a requirement for a partner VPN (private network) that required individual client machines to have PUBLIC addresses. >>As part of the VPN requirement we have set up a second set of DNS servers which are used to resolve hosts in the partner's domains. Why would you need separate DNS servers to handle this? ASB http://XeeMe.com/AndrewBaker <http://xeeme.com/AndrewBaker> Providing Virtual CIO Services (IT Operations & Information Security) for the SMB market... On Thu, Apr 24, 2014 at 8:26 AM, Melvin Backus <[email protected]> wrote: OK, this has been driving us nuts for a couple of days now. One of our remote sites is seeing seemingly random PCs change their DNS server settings. They're all configured to get them from the DHCP server, and it has the correct DNS servers. All the PCs do in fact get the correct settings when they get or renew an IP. That all seems to be working as we expect. But periodically we'll see a machine change the DNS servers to something else. This causes applications to start failing because the hosts they need no longer resolve. As soon as the PC renews it's IP, whether automatically or manually, everything goes back to normal and stuff works again. We have a short term fix (force the DNS server settings manually instead of DHCP) but that doesn't explain what's going on, and since we're using this same setup in 20 offices it also begs the question of why just this office. Background: Multiple small offices with either /28 or /27 networks. They are publicly routable IPs due to requirements for a partner VPN. The DHCP server is on the Juniper SSG FW. It servers two pools, one for PCs, another for phones. The PC subnet is publicly routable, the phone subnet is a non-routable 10.x subnet with matching ranges. (12.x.x.x/27 and 10.x.x.x/27). All DNS points to the home office. Until recently these pointed strictly to our domain DNS servers. As part of the VPN requirement we have set up a second set of DNS servers which are used to resolve hosts in the partner's domains. This is done with conditional forwarders. Partner DNS traffic gets resolved by their servers, everything else goes to our domain DNS or the Internet as required. This all works fine except in a single office. Even in that office it worked fine for weeks and has suddenly started this "revert" behavior. When the PCs change, they go back to pointing to our domain DNS which can't resolve the partner hosts. My question becomes (sorry it took so long) how do we track what is actually changing the DNS settings? I can tell when it happens fairly easily, but nothing in the event logs, etc., seems to indicate what triggered it, or what process is doing it. It doesn't happen as part of a DHCP operation as best we can tell. -------------------- Melvin Backus | Sr. Systems Analyst | Byers Engineering Company | 404.497.1565 Service Desk | 404-497-1599 | http://servicedesk.byers.com -- There are 10 kinds of people in the world... those who understand binary and those who don't.
