Okay, I need Linux Guru. I need to be able to monitor DHCP similarly to the monitoring that is possible in Windows. We are having an issue with DHCP and I am trying to narrow down exactly where the trouble lies. Below is a description of the problem.
Here is what I am dealing with.
On a network I support, (all windows servers) I have come across an extremely strange issue. I upgraded some old Compaq DL380 G1s to Dell 2850s. The new servers were given the same name and IP Address of the old Compaqs they were replacing. The problem arose almost immediately. Some, but not all, PCs were saying “Limited or no connectivity” and showed that they were not getting an IP address from the DHCP server. (The new Dells are now the Domain Controller, DNS Server, and DHCP Server at each location). Upon checking the DHCP scope on the server I found that the server HAD issued an IP to that client. I then discovered that if I hit F5 to refresh the view that the lease expiration would increment one second ever second. By doing packet captures simultaneously on the server and client I discovered that it appears that the initial broadcast is crossing Vlans. All switches are Cisco, and what’s more, some of the locations are in desperate need of upgrading. They are running equipment and CatOS/IOS that is about 6 years old. Other locations have been upgraded approximately 2 months ago and are running the latest and greatest of equipment and IOS. Here is some of the testing I’ve done:
1) Upgraded the drivers on both the server and client (three different versions on each)
2) Installed a 3Com 3c905 B NIC and the trouble disappeared. (This NIC is 10/100)
3) Installed 4 different gigabit Nics and the trouble remained.
4) Set the port speed, duplex, and flow control manually on the switches and Nics. No joy.
5) Moved a few of the PCs that were having issues into vlan 1 with the server so that no Vlan routing was taking place. Trouble disappeared.
6) Moved the DHCP scope onto the Cisco 6509 and problem disappeared.
So now it looks like the issue only occurs when I use 10/100/1000 Nics, and they must cross Vlans to acquire an IP.
We have since also installed 7 IBM 3650 at other locations. They are having the same issues.
I have now installed Redhat 9.0 onto one of the IBM’s to take Windows out of the equation. Dell and Cisco are both at a loss and are looking for any reason to blow me off. I need to confirm exactly where the trouble lies so that I can push for higher escalation of the issue.
Any ideas?
Here is what I am dealing with.
On a network I support, (all windows servers) I have come across an extremely strange issue. I upgraded some old Compaq DL380 G1s to Dell 2850s. The new servers were given the same name and IP Address of the old Compaqs they were replacing. The problem arose almost immediately. Some, but not all, PCs were saying “Limited or no connectivity” and showed that they were not getting an IP address from the DHCP server. (The new Dells are now the Domain Controller, DNS Server, and DHCP Server at each location). Upon checking the DHCP scope on the server I found that the server HAD issued an IP to that client. I then discovered that if I hit F5 to refresh the view that the lease expiration would increment one second ever second. By doing packet captures simultaneously on the server and client I discovered that it appears that the initial broadcast is crossing Vlans. All switches are Cisco, and what’s more, some of the locations are in desperate need of upgrading. They are running equipment and CatOS/IOS that is about 6 years old. Other locations have been upgraded approximately 2 months ago and are running the latest and greatest of equipment and IOS. Here is some of the testing I’ve done:
1) Upgraded the drivers on both the server and client (three different versions on each)
2) Installed a 3Com 3c905 B NIC and the trouble disappeared. (This NIC is 10/100)
3) Installed 4 different gigabit Nics and the trouble remained.
4) Set the port speed, duplex, and flow control manually on the switches and Nics. No joy.
5) Moved a few of the PCs that were having issues into vlan 1 with the server so that no Vlan routing was taking place. Trouble disappeared.
6) Moved the DHCP scope onto the Cisco 6509 and problem disappeared.
So now it looks like the issue only occurs when I use 10/100/1000 Nics, and they must cross Vlans to acquire an IP.
We have since also installed 7 IBM 3650 at other locations. They are having the same issues.
I have now installed Redhat 9.0 onto one of the IBM’s to take Windows out of the equation. Dell and Cisco are both at a loss and are looking for any reason to blow me off. I need to confirm exactly where the trouble lies so that I can push for higher escalation of the issue.
Any ideas?