SRX Cluster Fab Link Speeds and Reth Failover Times
[ Edited ]
We have an SRX1500 chassis cluster. It will be set up in an Active\Passive configuration eventually. Currently Active\Active but will revert. I have 2 questions of that's ok?
1. With regards to the fab links, I know that these can be 1G or 10G links. As we are going to be moving forward in an Active\Passive set up would I be right in saying that there would be absolutely no need for 10G link for the Fab links as there will be no transit traffic utilising these or is it possible that the non transit trafffic could possibly generate this sort of throughput?
As a side note I have not not configured anything for the heartbeat intervals and timers, I understand that there is a default but does not show when running the show chassis cluster heartbeat-interval\threshold command.
2. The initial Reth interface failover times in the cluster when testing are almost seamless when failing over from primary node to secondary node when I take a physical link down on the primary node child interface, when I bring the physical link back up on the primary node there is a 12 second loss of connectivity before traffic starts responding again. I suspect (may be wrong) that this may be to do with MAC address aging on the LAN between the chassis pair. Do I need to consider some type of gratuitous ARP setting on the cluster to update the MAC tables on the LAN? If you need any specific configuration supplied please let me know.