if the primary node of an HA config goes down, the controller stops responding

[Heather Lanigan]

Seen in 2.9.x and 3.0

Bootstrap lxd and enable-ha. Once settled, lxc stop the primary controller.

Looks like the leases aren't updating:

$ juju_leases
controller-leases:
  26c1ec8e-8c36-40a8-8b72-734ce860f066:
    holder: machine-0
    lease-acquired: 28s ago
    lease-expires: 32s
  312553fc-66ca-4f9b-8689-ce7ca76d1582:
    holder: machine-0
    lease-acquired: 21s ago
    lease-expires: 39s
  463126db-aa30-48ac-82c3-1e747fa0de04:
    holder: machine-0
    lease-acquired: 28s ago
    lease-expires: 32s
model-leases:
  463126db-aa30-48ac-82c3-1e747fa0de04:
    controller:
      holder: controller/0
      lease-acquired: 23s ago
      lease-expires: 37s

Over 10 minutes, the acquired and expires times do not change.

Recovers with a lxc start of the node stopped.

[Joseph Phillips (manadart)]

I believe this is resolved by:
https://github.com/juju/juju/pull/13429

[Joseph Phillips (manadart)]

I reopened this in error. I verified that it does not occur and the fix was released in the 2.9.17 milestone.