Fix the godoc for RollingUpdateDaemonSet to state that
spec.updateStrategy.rollingUpdate.maxUnavailable is rounded up.
A recent commit changed the godoc to say that the value of this field
was rounded down, but the actual implementation rounds up and always has
rounded up. (This is in contrast to Deployments, where
spec.strategy.rollingUpdate.maxUnavailable is rounded down.)
Follow-up to commit 5aa53f885c.
* api/openapi-spec/swagger.json:
* staging/src/k8s.io/api/apps/v1/generated.proto:
* pkg/apis/apps/types.go:
* staging/src/k8s.io/api/apps/v1/types.go:
* staging/src/k8s.io/api/apps/v1/types_swagger_doc_generated.go:
* staging/src/k8s.io/api/apps/v1beta2/generated.proto:
* staging/src/k8s.io/api/apps/v1beta2/types.go:
* staging/src/k8s.io/api/apps/v1beta2/types_swagger_doc_generated.go:
* staging/src/k8s.io/api/extensions/v1beta1/generated.proto:
* staging/src/k8s.io/api/extensions/v1beta1/types.go:
* staging/src/k8s.io/api/extensions/v1beta1/types_swagger_doc_generated.go:
* staging/src/k8s.io/cli-runtime/artifacts/openapi/swagger.json:
* staging/src/k8s.io/kubectl/testdata/openapi/swagger.json:
Change "rounding down" to "rounding up".
iSCSI and FC volume plugins do not implement real 3rd party attach/detach.
If reconstruction fails with an error on a FC or iSCSI volume, it will not
be unmounted from the volume global dir and at the same time it will be
marked as unused, to be available to be mounted on another node.
The volume can then be mounted on several nodes, resulting in volume
corruption.
The other block based volume plugins implement attach/detach that either
makes the volume stuck (can't be detached) or will be force-detached from a
node before attaching it somewhere else.
When UnmountDevice() of a FibreChannel volume fails after unmounting the
device and before the device is fully cleaned up, subsequent
UnmountDevice() retry won't find the device mounted and return without
retrying the device cleanup.
Therefore implement its own retry inside UnmountDevice() to make sure that
the volume devices are either fully cleaned or the error is serius enough
that even 1 minute of trying does not help.
Instead of listing all ReplicaSets in the namespace and checking their
controller UID, this patch adds a controllerUID index to the ReplicaSet
store and use it to get ReplicaSets with same controller, which reduces
the cost from O(#ReplicaSets) to O(1).
Benchmark results:
```
name old time/op new time/op delta
GetReplicaSetsWithSameController-48 18.2µs ± 9% 0.4µs ± 5% -97.64% (p=0.008 n=5+5)
name old alloc/op new alloc/op delta
GetReplicaSetsWithSameController-48 4.18kB ± 0% 0.05kB ± 0% -98.85% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
GetReplicaSetsWithSameController-48 15.0 ± 0% 2.0 ± 0% -86.67% (p=0.008 n=5+5)
```
When the StatefulSetMinReadySeconds feature gate is disabled,
the registry and validation must properly handle dropping the
minReadySeconds and AvailableReplicas fields
The minReadySeconds field on StatefulSet specifies the minimum
number of seconds for which a newly created Pod should be
ready without any of its containers crashing, for it to
be considered available. The AvailableReplicas field
in the status reflects the replicas that are available
PluginFactory is a function that returns a plugin. We have been
comparing these functions in unit tests and it has worked so far, but
starts to fail in gotip/master.
Note from the golang team:
```
Func values are incomparable. It is true that you could get the PC
through reflection but it is still not expected to be compared. PCs
can change due to inlining, wrappers, etc., depending on the
compiler's details, which is subject to change.
```
Signed-off-by: Davanum Srinivas <davanum@gmail.com>
