Currently if we disable cloud provider by the following
CLOUD_PROVIDER_FLAG=external
KUBE_FEATURE_GATES=DisableCloudProviders=true,DisableKubeletCloudCredentialProviders=true
we can no longer schedule workloads due to taints and a lack of
node configuration.
This pulls a CCM image from K/cloud-provider-gcp to run tests.
This is a pre-step for taking the above feature gates to beta.
It does not address the last known good dependency issue.
Specifically the CCM image is built on top of client-go and staging.
However this image will be an "old" verison of those libraries.
So it does not test if those libraries work in the CCM.
Fix shellcheck errors.
Add CCM_FEATURE_GATES for testing.
Switching to extended regex from perl regex.
Adding instrumentation to cluster configuration.
Improved regex to not greedily get key-value pairs.
Fixed issue with error on regex no line match.
Switch credentialprovider version to v1alpha1
Adding the ability to ignore no schedule flags in testing.
Specifically node.cloudprovider.kubernetes.io/uninitialized:NoSchedule
Fix shellcheck complaint.
Make GCS buckets created by the e2e tests regional.
There are by default multi-regional in US location. The change is
introduced to minimize GCS global cost.
GCS new pricing changed since Oct 1, 2022. See https://cloud.google.com/storage/pricing-announce
Signed-off-by: Arnaud Meukam <ameukam@gmail.com>
Add kube::util::wait-for-jobs in create-network() to wait
for background firewall rule creation, and if any fail, log
a message and exit with the failed job's return code.
COS-85 is very old OS image and is EoL in September 2022. Update to the
latest stable LTS version of COS.
Signed-off-by: David Porter <david@porter.me>
* Detect if image is using cgroupv2
* Configure both kubelet and containerd to use systemd cgroup driver
when running under cgroupv2
Systemd cgroup driver is recommended to be used when running on
cgroupv2. It is also the default in moby
https://github.com/moby/moby/pull/40846
Signed-off-by: David Porter <david@porter.me>
A previous commit updated the containerd version used on Windows
nodes from 1.5.4 to 1.6.2. However, the folder structure of the
containerd releases changed since then from:
cni/$binary.exe
to:
cni/bin/$binary.exe
Because of this, the Windows nodes do not have the necessary CNI
binaries needed to setup the pod networks.
containerd v1.6.0 introduced HostProcessContainers support [1], which
are required for e2e tests that need that feature.
This addresses some of the permafailing tests for Windows GCE E2E test runs.
[1] https://github.com/containerd/containerd/pull/5131
