I'm having an issue setting up two RB's redundant. The senario is:
ISP provides two internet connections and two cisco routers running HSRP.
These two ensures that if one internetconnection fail, the other will run. Or if one cisco router dies, the other will run. Hence the connection to ISP is redundant.
To make this work the ciscos has a build in 4 port switch where one port on each are used to connect them together.
I'll refer to the ciscos CISCO1 and CISCO2 from now on.
Behind this I want to connect two RB's redundant. I'll refer to them as RB1 and RB2.
The way I've done this is to configure a bridge called WAN including the ports eth3, 4 and 5. On this bridge I've enabled STP.
Eth5 on RB1 is connected to Eth5 on RB2.
Eth4 on RB1 is connected to CISCO1
Eth4 on RB2 is connected to CISCO2
Hence we have a switch-loop and the reason for having STP enabled: RB1.eth4 - CISCO1 - CISCO2 - RB2.eth4 - RB2.eth5 - RB1.eth5 - RB1.eth4
On the LAN side of the RB's:
Eth1 is set with Eth2 as a master-port.
RB1.eth1 is connected to RB2.eth1.
RB1.eth2 is connected to SWITCH1
RB2.eth2 is connected to SWITCH2
SWITCH1 is connected to SWITCH2
SWITCH1 and SWITCH2 has STP enabled, hence no need for STP enabled on the RB's on the LAN-side.
On eth2 I've created a VRRP-interface. Hence these routerboards work as master - standby.
Now, the issue:
This configuration seems to be working fine - for a few minutes. Then it stops working for 30-60 seconds. Then it starts working again for a few minutes.....
When it stops working the ping from RB1 to CISCO1 stop. So there is something about this link that is causing the issue.
Ping from any host on the LAN to the VRRP interface is stable.
So I suspected the STP on bridge 'WAN'. Hence I disconnected the cable between the ciscos - no change.
I tried RSTP instead of STP - no change
The only thing that makes the link stable is to pull the powercable on eigther one of the RB's.
Does anyone have any trick to make this stable and fully redundant?