Some tests use .yaml files to deploy pods, which have hardcoded
images. Those images cannot be used for Windows containers.
The image names can be injected by the tests themselves, based on
the configured registries.
By default, cmd is used to execute commands in Windows containers, but it
has some issues executing the command (no output):
echo 'hostName' | nc -w 1 -u <targetIP> <targetPort>
But it has no issues running (output is correct):
echo hostName | nc -w 1 -u <targetIP> <targetPort>
This causes the test "should function for node-pod communication: udp" to
fail when using Windows Containers.
Until we port the remaining e2e tests to use exec host pods instead
of SSH, it should be possible to use bastion host when
clusters don't expose public IPs for nodes. Testers can use the
KUBE_SSH_BASTION environment variable set to a `host:port` to
specify the host that the tests will use as a jump host.
If KUBE_SSH_BASTION is specified, the test will first connect to
the bastion using SSH and the provider private key, then establish
a tunnel from the bastion to the target node by its IP. If the
node has an external IP, the external IP must be routable from the
bastion.
Support KUBE_SSH_KEY_PATH as a single environment variable that if
specified ignores the provider specific settinsg and loads the
SSH private key from the provided path. Makes it much easier to
specify a consistent signer across providers.
Remove SLOW tag and update description for KUBEDESCRIBE(Probing container) and SIGDESCRIBE(EmptyDir Wrapper Volume)
Remove slow references for tests that execute below 5 minutes
Some tests which utilized SSH to run commands on nodes would
first look for external IPs but fall back to internal IPs
since those could be reachable by the testing program.
This change adds that same fallback logic to another method
used to find the appropriate SSH address for each node.
Fixes#68747
PR #70862 made each driver responsible for resetting its config, but
as it turned out, one place was missed in that PR: the in-tree gcepd
sets a node selector. Not resetting that caused other tests to fail
randomly depending on test execution order.
Now the test suite resets the config by taking a copy after setting up
the driver and restoring that copy before each test.
Long term the intention is to separate the entire test config from the
static driver info (https://github.com/kubernetes/kubernetes/issues/72288),
but for now resetting the config is the fastest way to fix the test flake.
Fixes: #72378
Exposing framework.VolumeTestConfig as part of the testsuite package
API was confusing because it was unclear which of the values in it
really have an effect. How it was set also was a bit awkward: a test
driver had a copy that had to be overwritten at test runtime and then
might have been updated and/or overwritten again by the driver.
Now testsuites has its own test config structure. It contains the
values that might have to be set dynamically at runtime. Instead of
overwriting a copy of that struct inside the test driver, the test
driver takes some common defaults (specifically, the framework pointer
and the prefix) when it gets initialized and then manages its own
copy. For example, the hostpath driver has to lock the pods to a
single node.
framework.VolumeTestConfig is still used internally and test drivers
can decide to run tests with a fully populated instance if needed (for
example, after setting up an NFS server).
This makes it possible to use the testsuites package out-of-tree
without pulling in unnecessary dependencies and code (in
test/e2e/storage/vsphere) that defines tests that are not wanted in a
custom test suite.
Different drivers support different volume sizes. Some have certain
minimum sizes, some maximum sizes. Instead of hard-coding some kind of
default into the testsuites, now each driver that supports dynamic
provisioning has to provide the size.
The setup of the V0 hostpath driver was done with copy-and-paste and
then changing just the driver name and the manifests. The same can be
achieved by making the base struct a bit more configurable, which
simplifies future changes (less code).
Renaming the provisioner container was unnecessary and was reverted to
make it possible to use the same patch configuration.
While at it, also fix the InitHostV0PathCSIDriver typo.
Commit 503f654d7a, "Update CSI tests to
point to 1.0.0 external bits", changed to images published on gcr.io,
perhaps because the images on quay.io weren't ready yet. We now have
up-to-date images on quay.io and should be using those.
Kubernetes Prow Robot