Add a Feature:HPA tag to these tests so they're not picked up by
the release-blocking job that focuses on [Serial] tests (but
excludes [Feature:.*] tests)
They take a combined 70 minutes on average. If they really need
to be in release-blocking as implemented, we should consider a
separate job to focus just on this feature.
Add a Feature:RegularResourceUsageTracking tag to these tests so
they're not picked up by the release-blocking job that focuses
on [Serial] tests (but excludes [Feature:.*] tests)
They take a combined 65 minutes on average. If they really need
to be in release-blocking as implemented, we should consider a
separate job to focus just on this feature.
HandleFlags() was used at e2e package and it depends on sub e2e
framework "config" in core e2e framework. That was invalid dependency.
So this moves HandleFlags() to e2e package for simple dependency.
We tried to separate logger functionality as subpackage of e2e test
framework, but we've recognized that should be core functionality
and we should keep it as core of e2e test framework after facing
circular dependency issues.
So this adds log.go back to core of e2e test framework. In addition,
this makes volume sub package use the core logger as a sample.
Also remove FS VolMode disruptive tests because they are already covered
through the subpath disruptive test cases.
Change-Id: Ice4b30b0d8fdcb1f7fd61e54d27f53557de9f13a
ConfirmStatefulPodCount() was used at e2e statefulset test only,
and that added dependency on another sub framework to statefulset
sub framework. This moves ConfirmStatefulPodCount() to the e2e test
for clean dependency.
Includes the changes from #80922 (that were reverted), and updates the test to appropriately add intolerable taints to all nodes except the target node
DesiredStateOfWorldPopulator should skip a volume that is not used in any
pod. "Used" means either mounted (via volumeMounts) or used as raw block
device (via volumeDevices).
Especially when block feature is disabled, a block volume must not get into
DesiredStateOfWorld, because it would be formatted and mounted there.
Every caller of ReadOrDie() specified ginkgo.Fail as fail argument,
and that was intentional to avoid depending on Ginkgo.
However that just spreaded the dependency on Ginkgo to caller sides.
Especially that was unnecessary e2e test framework "ingress" depended
on Ginkgo only for the above reason.
Now we are cleaning up the dependencies on e2e tests, so let's just
remove such dependencies.
GetGPUDevicePluginImage() was called in some e2e node test only.
So it is not worth keeping the function as a part of e2e test
framework. This moves GetGPUDevicePluginImage() to the e2e node
test for code cleanup.
In e2e test framework, ProxyRequest() is called in kubelet e2e test
framework only. In addition, the function works for kubelet as tests
which call the function do. So this moves the function into kubelet
e2e test framework to make the dependency simple.
Using a "normal" CSI driver for an inline ephemeral volume may have
unexpected and potentially harmful effects when the driver gets a
NodePublishVolume call that it isn't expecting. To prevent that mistake,
driver deployments for a driver that supports such volumes must:
- deploy a CSIDriver object for the driver
- set CSIDriver.Spec.VolumeLifecycleModes such that it contains "ephemeral"
The default for that field is "persistent", so existing deployments
continue to work and are automatically protected against incorrect
usage.
For the E2E tests we need a way to specify the driver mode. The
existing cluster-driver-registrar doesn't support that and also was
deprecated, so we stop using it altogether and instead deploy and
patch a CSIDriver object.
