Discussion Locked

Router/fwall cant talk to new server

0 helpful
5 replies
Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Hi,
I've faced this problem several times :
I have on my network Cyberguard firewall &
Cisco+Bay routers. When my server(web server
or VPN server) is replaced with a new box
which has exact configuration as the old server (ie same IP addr, disk, cpu etc), the
firewalls/routers can't seem to ping nor make tcp connections to the newly replaced box. The only change I can think off is that
this new box has a different MAC addr as it is using a different NIC. I have to reboot
the firewalls/routers & then they will start talking again. Someone told me this has to do with the ARP cache in the routers/firewal.

As my Cyberguard firewall is a cluster (High
Avail config) pair, whenever I switch the firewall cluster to the standby partner, I would find the new active firewall could not
communicate with the routers/server too
till I rebooted the routers/server.

Can someone explain & elaborate on this, if
possible email the replies to
gohpo@pacific.net.sg

I've always thought IP address is all that is needed to identify a node on the network
& we don't have to worry about ARP/MAC.

5 replies

by c king

Assuming that the new sever is operating properly (can other hosts on the same net ping it?), then the problem you have does sound like an ARP-cache problem.

When a firewall/router/PC/etc communicates with another device on the same LAN (Layer 2 network), they actually communicate via the MAC address (which is, oddly enough, a Layer 2 address). If all the firewall only knows the Layer 3 address of the recipient host (e.g. IP), then the firewall sends an ARP broadcast to find out what the recipient's MAC/hardware address is. When resolved, the Firewall creates a Layer 2 frame, which is then placed on the media (Ethernet, TR, etc.). To save devices from doing multiple lookups for the same device during a transmission stream, a cache iskept with already resolved Layer3 Addresses-to-MAC addresses. Your firewall's cache likely has a very long or infinite timeout on the cache, so when it attempts to connect to the server, it continues using the previous server's MAC address. The firewall reboot would clear this cache, as it is stored in RAM. The solution is to reconfigure your firewall with a shorter ARP-cache timeout (5 minutes is a safe bet - that way you only need to wait up to 5 minutes after switching boxes, avoiding a firewall reboot).

Theoretically, a intermediary switch could also cause this error. If it's MAC-address-table was corrupted, it could be forwarding the frames to the incorrect port. However, since a firewall reboot resolved the problem, it is highly unlikely that the switch is the problem device.

Good luck.

by ckilday

I agree with CKing about this being an arp cache problem. All computer devices communicate at layer 2 because this is the address that appears "on the wire". Layer 3 addressing provides for segmentation, scalability and the ability to communicate across different networks and media types through devices like routers and layer 3 switches.

Instead of performing a reboot every time you change a server, check your documentation for commands to clear or refresh your arp cache. For your cisco routers, you can enter the command "clear arp-cache" from the enable prompt and the router will clear its arp table and send out an arp request to every directly connected interface. This will rebuild your arp table much faster and without any noticeable loss of service like you get with a reboot. Your Bay routers and your firewall should have similar commands. Hope this helps.
- ck

This discussion is closed to new replies.