A recent test run failed in worker/provisioner/provisioner_test.go but passed before and since:
<http://juju-ci.vapour.ws:8080/job/github-merge-juju/900/console>
FAIL: provisioner_test.go:890: ProvisionerSuite.TestProvisionerObservesConfigChanges
...
[LOG] 0:00.666 INFO juju.provisioner provisioner-harvest-mode is set to destroyed; unknown instances not stopped []
[LOG] 0:00.666 INFO juju.provisioner stopping known instances [0xc210578a20]
[LOG] 0:00.666 INFO juju.provisioner removing dead machine "2"
[LOG] 0:00.666 DEBUG juju.apiserver <- [2A] machine-0 {"RequestId":43,"Type":"Provisioner","Request":"Remove","Params":{"Entities":[{"Tag":"machine-2"}]}}
provisioner_test.go:913:
// We default to the destroyed method. Only the one we know is
// dead should be stopped; not the unknown instance.
s.checkStopSomeInstances(c, []instance.Instance{i0}, []instance.Instance{i1})
provisioner_test.go:296:
c.Errorf("provisioner unexpectedly stopped instance %s", instId)
... Error: provisioner unexpectedly stopped instance dummyenv-1
...
I am raising this bug because it is failing multiple times in a day for ppc64el.