1/2
HA 2.0 in Sangfor HCI — Why Your VM Might NOT Fail Over (Even When You Expect It To)
  

George Fady Lv2Posted 2026-Sep-01 18:39

Hi everyone, I'm George, a Technical Support Engineer at Sangfor. I work daily on HCI tickets across the RMEA region, and I wanted to share something that trips up a lot of admins when they're first troubleshooting HA behavior.
The upgrade you might have missed: HA 1.0 → HA 2.0
Since HCI 6.10.0, Sangfor HA moved from a purely reactive model to a proactive one:

  • HA 1.0: waits for a VM crash or hardware failure, then restarts the VM

  • HA 2.0: predicts hardware degradation and platform anomalies, and can live-migrate a VM before it fails — not just restart it after

Two precautions that catch people out on tickets:

  • If you manually shut down a host from HCI Manager, that is not treated as a fault condition — HA will not trigger. This is expected behavior, not a bug.

  • VMs sitting on local (non-aSAN) storage won't fail over if the host dies — HA needs the VM's disk reachable from another node, which only works on distributed storage.

Discussion: Have you run into a case where HA didn't trigger and it turned out to be one of these two reasons? Or has HA 2.0's predictive migration actually saved you from an outage? Would love to hear real-world examples from the community.