MachinerSuite.TestMachinerMachineAssignedUnits fails intermittently

[Tim Penhey]

FAIL: machiner_test.go:162: MachinerSuite.TestMachinerMachineAssignedUnits

[LOG] 0:00.001 DEBUG juju.network no lxc bridge addresses to filter for machine
[LOG] 0:00.002 DEBUG juju.network "lxdbr0" has addresses [10.0.8.1/24 fe80::f44f:4dff:fe23:775f/64]
[LOG] 0:00.002 DEBUG juju.network not filtering address 255.255.255.255 for machine
[LOG] 0:00.002 DEBUG juju.network not filtering address 0.0.0.0 for machine
[LOG] 0:00.002 DEBUG juju.network addresses after filtering: [255.255.255.255 0.0.0.0]
[LOG] 0:00.003 INFO juju.worker.machiner setting addresses for machine-123 to ["255.255.255.255" "0.0.0.0"]
[LOG] 0:00.003 INFO juju.worker.machiner "machine-123" started
[LOG] 0:00.003 DEBUG juju.worker.machiner "machine-123" is now dying
machiner_test.go:182:
    // If EnsureDead fails with "machine has assigned units", then
    // the worker will not fail, but will wait for more events.
    c.Check(err, jc.ErrorIsNil)
... value *errors.Err = &errors.Err{message:"agent should be terminated", cause:error(nil), previous:error(nil), file:"github.com/juju/juju/worker/errors.go", line:20} ("agent should be terminated")
... error stack:
 github.com/juju/juju/worker/errors.go:20: agent should be terminated

OOPS: 13 passed, 1 FAILED
--- FAIL: TestPackage (24.32s)

[Anastasia (anastasia-macmood)]

We have made a lot changes since this failure was reported. I can no longer reproduce it locally nor under stress.