F5-301A: High Availability and Device Service Clustering

Concepts

High availability rests on three layers. A device trust establishes secure peer communication using certificates. A device group defines which configuration is synchronised; sync-failover groups also control failover, sync-only groups do not. A traffic group owns the floating objects - floating self IPs, virtual server addresses and, optionally, a MAC masquerade address - and moves them to the active device.

Failover detection can use network failover (heartbeats over the network, unicast or multicast) or hardwired failover (a direct cable). Configuration sync and failover are independent: synchronising configuration does not trigger failover, and connections are not mirrored unless connection mirroring is enabled on the virtual servers.

Active/active requires additional design work - MAC masquerade, SNAT, or routing that tolerates two devices answering.

Traps

• Assuming a failover event copies configuration. It does not; run config sync.
• Expecting sessions to survive failover without connection mirroring.
• Mismatched NTP, VLANs or self IPs between peers, which blocks the trust or the failover.
• Expecting active/active to work without distinct addressing or routing design.
• Forgetting that config sync takes the running configuration from the device on which you initiate it.
• Floating objects left out of the traffic group, so they never move.

Worked example

Two devices, one floating address:

1. Create the device trust from device A, adding B's management address and credentials.
2. Create a sync-failover device group containing both devices.
3. Create a traffic group and add the floating self IP to it.
4. Perform initial sync from the device you want as the source of truth.
5. Test with tmsh run sys failover standby on the active unit and confirm the floating address still answers.

Rule of thumb: trust → device group → traffic group → initial sync → failover test. If a floating object does not move, it is not in the traffic group.