The k3s distribution of k8s uses an embedded version of containerd and
configures it to log to a file, not the journal. Hence, although we
collect the journal as a test artifact, we also need to collect the
actual log files for containerd.
Also collect the k3s containerd config files to help with debugging.
Fixes: #9104.
Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
The jobs running on garm will collect journal information. The data gathered
is based on the time the tests started running. The $start_time is
exported on run_tests() and used in collect_artifacts(). It happens that
run_tests() and collect_artifacts() are called on different steps of the
workflow and the environment variables aren't preserved between them,
i.e, $start_time exported on the first step is not available on the
subsequents.
To solve that issue, let's save $start_time in the file pointed out by
$GITHUB_ENV that Github actions uses to export variables. In case $GITHUB_ENV is
empty then probably it is running locally outside of Github, so it won't
save the start time value.
Fixes#9217
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Until this point the deployed KBS service is only reachable from within
the cluster. This introduces a generic mechanism to apply an Ingress
configuration to expose the service externally.
The first implemened ingress is for AKS. In case the HTTP application
routing isn't enabled in the cluster (this is required for ingress), an
add-on is applied.
It was added the get_cluster_specific_dns_zone() and
enable_cluster_http_application_routing() helper functions
to gha-run-k8s-common.sh.
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Introduce the tests/integration/kubernetes/confidential_kbs.sh library
that contains functions to manage the KBS on CI. Initially implemented
the kbs_k8s_deploy() and kbs_k8s_delete() functions to, respectively,
deploy and delete KBS on Kubernetes. Also hooked those functions in the
tests/integration/kubernetes/gha-run.sh script to follow the convention
of running commands from Github Workflows:
$ .tests/integration/kubernetes/gha-run.sh deploy-coco-kbs
$ .tests/integration/kubernetes/gha-run.sh delete-coco-kbs
Fixes#9058
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
This PR adds the collect artifacts function in gha-run script for
the kubernetes tests.
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
Currently, the checking for kata-deploy is running assume that the
daemonset scheduled at least one pod, however it might not had and the
kubectl wait command fails due to "error: no matching resources found".
On CI I've observed that fail intermittently. I suspect the service
account kata-deploy-sa take a while to show up then no kata-deploy is
scheduled in meanwhile.
Changed the checker logic to use waitForProcess() to keep testing if it is
already running, or hit the timeout (still 10m).
Fixes#9183
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Kubernetes v1.29 introduced a new `PodReadyToStartContainers` condition
that gets inserted at index 0 in the conditions array. This means that
the expected `PodCompleted` reason can now be either at index 0 with
kubernetes v1.28 and older or at index 1 starting with kubernetes v1.29.
This is fragile at best since the `kubectl wait` doesn't allow to combine
multiple checks. Also, checking the reason is dubious as it doesn't really
tell if the pods have actually completed or not.
Check the pod phase to be `Succeeded` instead, this guarantees that :
> All containers in the Pod have terminated in success, and will not
> be restarted.
Fixes#9178
Signed-off-by: Greg Kurz <groug@kaod.org>
Changed the "run k8s tests on AKS" workflows to get the CoCo KBS
installed so that we can run attestation tests.
The plan is to run attestation tests only on a subset of non-TEE jobs
initially, so this commit restricts to install KBS only on kata-qemu
configuration. Actually at this point it is added only stubs commands
to tests/integration/kubernetes/gha-run.sh that should be implemented
in a future commit.
Fixes#9058
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
CI failed to deploy nydus snapshotter because it was not cleaned up last time.
So we can try to cleanup nydus snapshotter before deploying it.
Fixes: #9121
Signed-off-by: ChengyuZhu6 <chengyu.zhu@intel.com>
Co-authored-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
teardown() gets executed after each test case, so there is no need to
clean-up before teardown.
Fixes: #9072
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Add set_namespace_to_policy_settings() for changing the pod namespace
in genpolicy settings.
Fixes: #9072
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
This has been introduced by 53bc4a432b,
where the condition was changed.
The correct condition is:
* If the list of supported tees does not contain the kata hypervisor
and the list of supported non tees does not contain the kata
hypervisor.
The error is that we were checking whether kata-hypervisor would contain
the list of supported tees, and that would almost always be false
(unless in the case where the list had an one and only one element).
Fixes: #9055 -- part II
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
A test `vcpu allocation k8s test` exhibits different behavior on s390x
For more details, please refer to issue #9093.
This commit is to make the test skipped until the issue is resolved on
the platform.
Fixes: #9093
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
This PR adds the ability to run k8s confidential tests in a
non-TEE environment.
Fixes#9055
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
This will ensure no leftovers are in the node, which has been cause the
TDX CI to fail every now and then.
Fixes: #9081
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Delete the debugger pod created during the test, rather than already
existing debugger pods.
Also, send the output of "kubectl delete" to stderr, just in case it's
useful for debugging.
Fixes: #9069
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
We can use daemonset to deploy nydus snapshotter, which will decrease
one manual step both for Kata Containers and Confidential Containers CI.
Fixes: #8584
Signed-off-by: ChengyuZhu6 <chengyu.zhu@intel.com>
This PR replaces the add_kernel_initrd_annotations_to_yaml function
more generic so later can be used for other components.
Fixes#9054
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
1. add_requests_to_policy_settings allows one or more ttrpc requests
from the Host to the Guest. Example:
add_requests_to_policy_settings "${policy_settings_dir}" \
"ReadStreamRequest" "WriteStreamRequest"
2. add_copy_from_host_to_policy_settings allows executing on the Guest
the commands initiated behind the scenes by "kubectl cp" from the
Host to the Guest. Example:
add_copy_from_host_to_policy_settings "${policy_settings_dir}"
3. add_copy_from_guest_to_policy_settings allows executing on the Guest
the commands initiated behind the scenes by "kubectl cp" from the
Guest to the Host. Example:
add_copy_from_guest_to_policy_settings "${policy_settings_dir}" \
"/tmp/file.txt"
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
1. Rename install_kata_common to install_kata_core.
2. Add TODO for better way to install the Kata tools.
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Automatically generate the test policy for k8s-attach-handlers.bats,
if AUTO_GENERATE_POLICY is enabled.
Steps:
- Create a temporary directory for the current test and copy the
common genpolicy settings into this new directory.
- Change genpolicy settings in the temp directory to allow the
"kubectl exec" command that this test needs. (For CoCo, exec is
blocked by the default policy settings)
- Auto-generate the policy for the test YAML file.
- Test as usual, using the YAML file.
- Clean-up the temporary settings described above.
Fixes: #8921
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Enable AUTO_GENERATE_POLICY for one of the Kata CI K8s test platforms.
Additional platforms will be enabled after testing them.
When AUTO_GENERATE_POLICY is enabled, create genpolicy settings that
are common for all tests. Some of the tests will make temporary copies
of these common settings and customize them as needed.
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
1. Avoid repeating "kata-containers-k8s-tests".
2. Allow users to specify a different test namespace.
3. Introduce the TEST_CLUSTER_NAMESPACE variable, that will also be
useful when auto-generating the Agent Policy for these tests.
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
This PR skips the k8s tests that are not working with cloud hypervisor
runtime-rs with its proper issue.
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
This PR is a split of #8585.
make the changes on the Github workflows, and the skeleton to deploy_snapshotter()
and cleanup_snapshotter() in tests/integration/kubernetes/gha-run.sh in this commit.
After initially merging this patch to trigger CI jobs for CoCo, which will begin executing
the dummy functions deploy_snapshotter() and cleanup_snapshotter(), the implementation details for these functions
remain in #8585. Our subsequent step involves transferring this logic to the PR #8484, enabling the PR to undergo CI testing prior to its merge.
Fixes: #8997
Signed-off-by: ChengyuZhu6 <chengyu.zhu@intel.com>
This PR adds workflow for running kubernetes test suite on ppc64le.
It uses scripts to create and delete the cluster using kubeadm as none of the current cluster creation tools are supported on Power.
Fixes: #7950
Signed-off-by: Amulyam24 <amulmek1@in.ibm.com>
The existing confidential basic test titled `Test unencrypted
confidential container launch success and verify that we are
running in a secure enclave` has been updated to incorporate
IBM Secure Execution (`qemu-se`).
Previously, a secure image was absent from kata-deploy, hindering
the inclusion of IBM SE in the test.
Thanks to the #6755 update, it is now possible to test the TEE.
This modification extends the existing test by introducing
`qemu-se`. The specific changes are outlined below:
- Add an additional test `cc-se-e2e-tests` to s390x nightly
- Expansion of `REMOTE_COMMAND_PER_HYPERVISOR` for `qemu-se`
- Temporary exclusion of two test cases currently incompatible with IBM SE
(`cpu-ns` is a common issue across all TEEs, while `inotify`
will be addressed in a subsequent pull request).
Fixes: #8913
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
Add --show-output-of-passing-tests to the k8s integration tests. The
output of a passing test can be helpful when investigating a failure
of the same test.
Fixes: #8885
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
The changes to install and test genpolicy must come later, after CI
picks up these gha changes.
Fixes: #8856
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Log the list of the current pods between tests because these pods
might be related to cluster nodes occasionally running out of memory.
Fixes: #8769
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
To become more resilient against these kinds of errors:
deployment.apps/confidential-unencrypted created
pod/confidential-unencrypted-c5fdd6964-rrb6q condition met
ssh: connect to host 10.42.0.109 port 22: Connection refused
Fixes: #8687
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
This PR fixes the indentation of the confidential common
script for kubernetes tests.
Fixes#8698
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
This will make the life easier for dragonball developers to properly
enable the tests once the tests are ready.
Fixes: #8569
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Print more information useful for debugging. Also, use a separate YAML
file for this test, instead of reusing someone else's file.
Fixes: #8270
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Implements the following test case:
Scenario: Check incorrect hash fails
**Given** I have a version of kata installed that has a kernel with the
initramfs built and config with rootfs_verity.scheme=dm-verity
rootfs_verity.hash=<incorrect hash of rootfs> set in the kernel_params
**When** I try and create a container a basic pod
**Then** The pod is doesn't run
**And** Ideally we'd get a helpful message to indicate why
Currently on CI only qemu-tdx is built with measured
rootfs support in the kernel, so the test is restriced to that
runtimeclass.
Fixes#7415
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Bring the setup_common() from CCv0 branch test's
integration/kubernetes/confidential/tests_common.sh. It should be used
to reduce boilerplates on the setup() of the tests.
Unlike the original code, this won't export the `test_start_time` variable
as it wouldn't be accurate to grab logs from the worker nodes due
date/time mismatch between the running tests machine and the worker
node. The function export the `node` variable which holds the name of
a random node which has kata installed. Apart from that, it exports the
`node_start_time` which capture the date/time when the test started,
relative to the `node`.
Tests that should inspect the logs can schedule pods/resources to the `node`
and use `node_start_time` as the value reference to grep the logs.
Fixes#7590
Co-authored-by: stevenhorsman <steven@uk.ibm.com>
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Bring the assert_logs_contain() from CCv0 branch tests'
integration/kubernetes/confidential/lib.sh.
Introduced the print_node_journal() which uses `kubectl debug` to print
the systemd's journal of a k8s's node.
Fixes#7590
Co-authored-by: stevenhorsman <steven@uk.ibm.com>
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
This new function allow to the annotations to metadata section in a yaml
configuration file.
Co-authored-by: Ryan Savino <ryan.savino@amd.com>
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Copied the new_pod_config() and pod-config.yaml.in from CCv0 branch
tests' integration/kubernetes/confidential/tests_common.sh and fixtures.
Unlike the original version, new_pod_config() now gets the runtimeclass
by parameter as the RUNTIMECLASS environment variable seems not broadly
used on main branch's CI.
The pod-config.yaml.in was changed as the diff shows below. In
particular the imagePullSecrets was removed to avoid it throwing a
warning on the pod's log.
```
--- a/tests/integration/kubernetes/runtimeclass_workloads/pod-config.yaml.in
+++ b/tests/integration/kubernetes/runtimeclass_workloads/pod-config.yaml.in
@@ -5,12 +5,10 @@
apiVersion: v1
kind: Pod
metadata:
- name: busybox-cc
+ name: test-e2e
spec:
runtimeClassName: $RUNTIMECLASS
containers:
- - name: nginx
+ - name: test_container
image: $IMAGE
- imagePullPolicy: Always
- imagePullSecrets:
- - name: cococred
\ No newline at end of file
+ imagePullPolicy: Always
\ No newline at end of file
```
Co-authored-by: Georgina Kinge <georgina.kinge@ibm.com>
Co-authored-by: Megan Wright <Megan.Wright@ibm.com>
Co-authored-by: stevenhorsman <steven@uk.ibm.com>
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The following functions were copied from CCv0's branch test's
integration/kubernetes/confidential/lib.sh. I did just smalls
refactorings (shortened their names and delinted shellcheck warnings):
- k8s_delete_all_pods_if_any_exists()
- k8s_wait_pod_be_ready()
- k8s_create_pod()
- assert_pod_fail()
Co-authored-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Co-authored-by: Georgina Kinge <georgina.kinge@ibm.com>
Co-authored-by: Jordan Jackson <jordan.jackson@ibm.com>
Co-authored-by: Megan Wright <Megan.Wright@ibm.com>
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Co-authored-by: Wang, Arron <arron.wang@intel.com>
As part of the CI migration, this PR is to add workflows for containerd and k8s for s390x.
Fixes: #7930
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
As we've done some changes in the VMM vcpu allocation, let's introduce
basic tests to make sure that we're getting the expected behaviour.
The test consists in checking 3 scenarios:
* default_vcpus = 0 | no limits set
* this should allocate 1 vcpu
* default_vcpus = 0.75 | limits set to 0.25
* this should allocate 1 vcpu
* default_vcpus = 0.75 | limits set to 1.2
* this should allocate 2 vcpus
The tests are very basic, but they do ensure we're rounding things up to
what the new logic is supposed to do.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
The current kata-deploy code has been doing a `sed` to add allowed
hypervisor annotations, so CBL mariner can be tested with their own
kernel and initrd.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Encode policy file during test - easier to understand than hard-coding
the encoded file contents.
Fixes: #8214
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
The KUBERNETES variable is mostly used by kata-deploy whether to apply
k3s specific deployments or not. It is used to select the type of
kubernetes to be installed (k3s, k0s, rancher...etc) and it is always
set on CI. Running the script locally we want to set a value by default
to avoid `KUBERNETES: unbound variable` errors.
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
This test can give false-positive on a multi-node cluster. Changed it to
use the new get_one_kata_node() and the modified exec_host() to run the
setup commands on a given node (that has kata installed) and ensure the
test pod is scheduled at that same node.
Fixes#7619
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
This test can give false-positive on a multi-node cluster. Changed it to
use the new get_one_kata_node() and the modified exec_host() to run the
setup commands on a given node (that has kata installed) and ensure the
test pod is scheduled at that same node.
Fixes#7619
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The exec_host() simply fails on cluster with multi-nodes because
`kubectl get node -o name" will return a list o names. Moreover, it will
return control nodes names which usually don't have kata installed.
Fixes#7619
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The introduced get_one_kata_node() returns the first node that
has the kata-runtime=true label, i.e., supposedly a node with
kata installed.
This is useful for tests that should run on a determined worker
node on a multi-nodes cluster.
Fixes#7619
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Let KATA_HYPERVISOR be qemu by default in gh-run.sh as this variable
is required to tweak some configurations of kata-deploy.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The deploy-kata() of gha-run.sh will wait for 10 minutes for the kata
deploy installation finish. This allow users of the script to overwrite
that value by exporting the KATA_DEPLOY_WAIT_TIMEOUT environment
variable.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Fixed a couple of warns shellcheck emitted and disabled others:
* SC2154 (var is referenced but not assigned)
* SC2086 (Double quote to prevent globbing and word splitting)
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The only difference to the other platforms is that it needs to
export KUBECONFIG.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The cleanup-kcli() behaves like other deploy kata for
bare-metal (e.g. sev, tdx...etc) except that KUBECONFIG
should be exported.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The cleanup-kcli() behaves like other clean up for bare-metal (e.g. sev,
tdx...etc) except that KUBECONFIG should be exported.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
On CI workflows the variables DOCKER_REGISTRY, DOCKER_REPO and
DOCKER_TAG are exported to match the built image. However, when running
the script outside of CI context, a developer might just use the latest
image which in this case will be
`quay.io/kata-containers/kata-deploy-ci:kata-containers-latest`.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Adapted the gha-run.sh script to create a Kubernetes cluster locally
using the kcli tool.
Use `./gha-run.sh create-cluster-kcli` to create it, and
`./gha-run.sh delete-cluster-kcli` to delete.
Fixes#7620
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
The k8s.gcr.io is deprecated for a while now and has been redirected to
registry.k8s.io. However on some bare-metal machines in our testing
pools that redirection is not working, so let's just replace the
registries.
Fixes#8098
Signed-off-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
(cherry picked from commit b2c3bca558c38deff2117d5909d9071c23c05590)
We've faced this as part of the CI, only happening with the CRI-O tests:
```
not ok 1 Test readonly volume for pods
# (from function `exec_host' in file tests_common.sh, line 51,
# in test file k8s-file-volume.bats, line 25)
# `exec_host "echo "$file_body" > $tmp_file"' failed with status 127
# [bats-exec-test:38] INFO: k8s configured to use runtimeclass
# bash: line 1: $'\r': command not found
#
# Error from server (NotFound): pods "test-file-volume" not found
```
I must say I didn't dig into figuring out why this is happening, but we
may be safe enough to just trail the '\r', as long as all the tests keep
passing on containerd.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
This is based on official CRI-O documentations[0] and right now we're
making this specific to Ubuntu as that's what we have as runners.
We may want to expand this in the future, but we're good for now.
[0]:
https://github.com/cri-o/cri-o/blob/main/install.md#apt-based-operating-systems
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Otherwise only the first test will be executed
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let me start with a fair warning that this commit is hard to split into
different parts that could be easily tested (or not tested, just
ignored) without breaking pieces.
Now, about the commit itself, as we're on the run to reduce costs
related to our sponsorship on Azure, we can split the k8s tests we run
in 2 simple groups:
* Tests that can be run in the smaller Azure instance (D2s_v5)
* Tests that required the normal Azure instance (D4s_v5)
With this in mind, we're now passing to the tests which type of host
we're using, which allows us to select to run either one of the two
types of tests, or even both in case of running the tests on a baremetal
system.
Fixes: #7972
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
The list of tests which require a bigger VM instance is:
* k8s-number-cpus.bats -- failing on all CIs
* k8s-parallel.bats -- only failing on the cbl-mariner CI
* k8s-scale-nginx.bats -- only failing on the cbl-mariner CI
We'll keep those disabled while we re-work the logic to **only run
those** in a bigger (and more expensive) VM instance.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Without setting the cpu limit / request to 1, we can make this test run
in a smaller VM instance without any issue.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Fix kernel and initrd annotations in the k8s tests on Mariner. These
annotations must be applied to the spec.template for Deployment, Job
and ReplicationController resources.
Fixes: #7764
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
There's absolutely no need to have the skip check as part of the test
itself when it's already done as part of the setup function.
We're only touching the files here that were touched in the previous
commit.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's keep both checks for now, but in the future we'll be able to
remove the check for "firecracker", as the hypervisor name used as part
of the GitHub Actions has to match what's used as part of the
kata-deploy stuff, which is `fc` (as in `kata-fc for the runtime class)
instead of `firecracker`.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
We've been using the `kata-deploy-tdx` target as that also uses k3s as
base, but it's better to just have a specific garm target.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
So we have a better control on which flavour of kubernetes kata-deploy
is expected to be targetting.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
As we were using `tee` without the `-a` (or `--apend`) aptton, the
containerd config would be overwritten, leading to a NotReady state of
the Node.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's download the vanilla kubectl binary into `/usr/bin/`, as we need
to avoid hitting issues like:
```sh
error: open /etc/rancher/k3s/k3s.yaml.lock: permission denied
```
The issue basically happens because k3s links `/usr/local/bin/kubectl`
to `/usr/local/bin/k3s`, and that does extra stuff that vanilla
`kubectl` doesn't do.
Also, in order to properly use the k3s.yaml config with the vanilla
kubectl, we're copying it to ~/.kube/config.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Otherwise the /etc/rancher/k3s/k3s.yaml is not readable by other users
than root.
As --write-config-mode is being passed, and that's an option that has to
be passed to the `server`, -s is also added to the command line.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
`wait` waits for a job to complete, not a number of seconds. Not sure
how I got that wrong in the first place, but it's what it's.
Fixes: #6542
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
This function right now is completely based on what's part of the tests
repo[0], and that's the reason I'm keeping the `Signed-off-by` of all
the contributors to that file.
This is not perfect, though, as it changes the default snapshotter to
devmapper, instead of only doing so for the Kata Containers specific
runtime handlers. OTOH, this is exactly what we've always been doing as
part of the tests.
We'll improve it, soon enough, when we get to also add a way for
kata-deploy to set up different snapshotters for different handlers.
But, for now, this is as good (or as bad) as it's always been.
It's important to note that the devmapper setup doesn't take into
consideration a BM machine, and this is not suitable for that. We're
really only targetting GHA runners which will be thrown away after the
run is over.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Signed-off-by: Shiming Zhang <wzshiming@foxmail.com>
Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
Signed-off-by: Gabriela Cervantes <gabriela.cervantes.tellez@intel.com>
One can use different kubernetes flavours for getting a kubernetes
cluster up and running.
As part of our CI, though, I really would like to avoid contributors
spending time maintaining and updating kubernetes dependencies, as done
with the tests repo, and which has been proven to be really good on
getting things rotten.
With this in mind, I'm taking the bullet and using "k3s" as the way to
deploy kubernetes for the devmapper related tests, and that's the reason
I'm adding a function to do so, and this will be used later on as part
of this series.
It's important to note that the k3s setup doesn't take into
consideration a BM machine, and this is not suitable for that. We're
really only targetting GHA runners which will be thrown away after the
run is over.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Use AGENT_POLICY=yes when building the Guest images, and add a
permissive test policy to the k8s tests for:
- CBL-Mariner
- SEV
- SNP
- TDX
Also, add an example of policy rejecting ExecProcessRequest.
Fixes: #7667
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Let's expand the confidential test to also support TDX.
The main difference on the test, though, is that we're not grepping for
a string in the `dmesg` output, but rather relying on `cpuid` to detect
a TDX guest.
Fixes: #7184
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Add a test case for the launch of unencrypted confidential
container, verifying that we are running inside a TEE.
Right now the test only works with SEV, but it'll be expanded in the
coming commits, as part of this very same series.
Fixes: #7184
Signed-Off-By: Unmesh Deodhar <udeodhar@amd.com>
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
k8s-pid-ns.bats was already using the test name from
k8s-kill-all-process-in-container.bats - probably a copy/paste bug.
Fixes: #7753
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
By doing this we can make sure there won't be any clash on the cluster
name created for either the k8s or the kata-deploy tests.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
In a follow-up series, we'll add a whole suite for the kata-deploy
tests. With this in mind, let's already get rid of this one and avoid
more kata-deploy tests to land here.
Fixes: #7642
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
The default `kata` runtime class would get created with the `kata`
handler instead of `kata-$KATA_HYPERVISOR`. This made Kata use the wrong
hypervisor and broke CI.
Fixes: #7663
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
Let's split a good portion of `tests/integration/kuberentes/gha-run.sh`
out, and put them in a place where they can be used to the soon-to-come
kata-deploy specific tests.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
With these 2 simple checks we can ensure that we do not regress on the
behaviour of allowing the runtime classes / default runtime class to be
created by the kata-deploy payload.
Fixes: #7491
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's add here the image we'll be using for unencrypted confidential
tests. Later on, we'll make sure to build and use this image as part of
our CI.
The image can easily be built as a multi-arch image, and has `cpuid`
installed in case of `x86_64` build, so it can be used to detect whether
we're running on a TEE guest without having to rely on `dmesg | grep
...`.
Fixes: #7595
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
We don't have to do any sed to replace the runtimeclass being used by
the moment we start taking advantage of the `DEFAULT_SHIM` environment
variable exposed merged in the previous commits.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Instead of using package manager to install bats, building
this from source. This gives us the updated version of bats
which supports functions such as setup_file and
teardown_file.
We can use these functions into our current tests.
Fixes: #7597
Signed-off-by: Unmesh Deodhar <udeodhar@amd.com>
Otherwise the VM deletion may not delete, leaving us with several
machines behind.
Fixes: #7509
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
The amount of info we've added seemed unnecessary, and ends up making
our lives even harder when trying to find errors.
Let's just rely on the kata-debug container to collect the needed info
for us.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
It's been proven to not be useful, and ends up making things more
confusing due to the amount of logs printed.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's make sure we can debug kata-deploy in case something goes wrong
during its execution.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
This can be easily done as there was no official release with the
previous values.
The reason we're doing so is because when using `yq` to replace the
value, even when forcing `--tag '!!str' "yes"`, the content is placed
without quotes, causing errors in our CI.
While here, we're also removing the fallback value for DEBUG, as it is
**always** set in the kata-deploy.yaml file.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
This imports the k8s-file-volume test from the tests repo and modifies
it slightly to set up the host volume on the AKS host.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This imports the k8s-volume test from the tests repo and modifies it
slightly to set up the host volume on the AKS host.
Fixes: #6566
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This deletes node debugger pods after execution since their presence may
affect tests that assume only test workloads pods are present.
For example, in `k8s-job` we wait for *any* pod to be in the `Succeeded`
state before proceeding, which causes failures.
Fixes: #7452
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This simply allows setting a custom resource group when debugging
locally, so as to prevent name collisions and not pollute the namespace.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
Makes it so that `setup.sh` doesn't make changes in
`runtimeclass_workloads/` directly. Instead we treat that as a template
directory and we use the new directory `runtimeclass_workloads_work/` as
a work dir.
This has two advantages:
* Allows rerunning tests without the assumption that `setup.sh` must be
idempotent. E.g. the `set_runtime_class()` step would break.
* Doesn't pollute your git environment with a bunch of changes when
developing.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This splits deploying Kata and running the tests into separate commands
to make it possible to rerun tests locally without having to redeploy
Kata each time.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This is already declared as part of the `common.bash` file, so let's
just make sure we use it from there.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
This will help us to in two fronts:
* catching possible issues related to kata-deploy cleanup
* do more (like, in the future, collect logs) after the tests run
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Currently a mixture of cbl-mariner and mariner is used when creating the
mariner initrd. The kata-static tarball has mariner in the name, but the
jenkins url uses cbl-mariner. This breaks cache usage.
Use mariner as the target name throughout the build, so that caching works.
Fixes: #7292
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
We've noticed this caused regressions with the k8s-oom tests, and then
decided to take a step back and do this in the same way it was done
before 67972ec48a.
Moreover, this step back is also more reasonable in terms of the
controlling logic.
And by doing this we can re-enable the k8s-oom.bats tests, which is done
as part of this PR.
Fixes: #7271
Depends-on: github.com/kata-containers/tests#5705
Signed-off-by: Yushuo <y-shuo@linux.alibaba.com>
Let's skip the k8s-oom, as the test is currently failing.
We've an issue opened for that, and we'll be working on re-enabling it
as soon as possible.
Reference:
https://github.com/kata-containers/kata-containers/issues/7271Fixes: #7253
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's skip the k8s-number-cpus, as the test is currently failing.
We've an issue opened for that, and we'll be working on re-enabling it
as soon as possible.
Reference:
https://github.com/kata-containers/kata-containers/issues/7270Fixes: #7253
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
Let's make sure we run our tests in a specific namespace, as in case of
any kind of issue, we will just get rid of the namespace itself, which
will take care of cleaning up any leftover from failing tests.
One important thing to mention is why we can get rid of the `namespace:
${namespace}` on the tests that are already using it, and let's do it in
parts:
* namespace: default
We can easily get rid of this as that's the default namespace where
pods are created, so it was a no-op so far.
* namespace: test-quota-ns
My understanding is that we'd need this in order to get a clean
namespace where we'd be setting a quota for. Doing this in the
namespace that's only used for tests should **not** cause any
side-effect on the tests, as we're running those in serial and there's
no other pods running on the `kata-containers-k8s-tests` namespace
Last but not least, we're not dynamically creating namespaces as the
tests are not running in parallel, **never**, not in the case of having
2 tests being ran at same time, neither in the case of having 2 jobs
being scheduled to the same machine.
Fixes: #6864
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
We only need to install in run_tests() so that the yq install is picked up by
kubernets/setup.sh as well. We also need to either use (sudo &&
INSTALL_IN_GOPATH=false) || (INSTALL_IN_GOPATH=true).
Signed-off-by: Jeremi Piotrowski <jpiotrowski@microsoft.com>
Mariner ships a bleeding-edge kernel that might be ahead of upstream, so
we use that to guarantee compatibility with the host.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
* Adds a new `rootfs-initrd-mariner` build target.
* Sets the custom initrd path via annotation in `setup.sh` at test
time.
* Adapts versions.yaml to specify a `cbl-mariner` initrd variant.
* Introduces env variable `HOST_OS` at deploy time to enable using a
custom initrd.
* Refactors the image builder so that its caller specifies the desired
guest OS.
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
The current testing setup only supports running Kata on top of an Ubuntu
host. This adds Mariner to the matrix of testable hosts for k8s
tests, with Cloud Hypervisor as a VMM.
As preparation for the upcoming PR that will change only the actual test
code (rather than workflow YAMLs), this also introduces a new file
`setup.sh` that will be used to set host-specific parameters at test
run-time.
Fixes: #6961
Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
Currently Kata does not support memory / CPU hotplug for SEV or
SEV-SNP so we need to skip tests that rely on it.
Signed-off-by: Tobin Feldman-Fitzthum <tobin@ibm.com>
Now that SEV artifacts are built by GHA, remove
conditional that skips tests when using qemu-sev.
Signed-off-by: Tobin Feldman-Fitzthum <tobin@ibm.com>
Now that we have SNP artifacts in place and they are built via gha,
remove the condition that skips the tests for SNP.
Fixes: #6809
Signed-off-by: Tobin Feldman-Fitzthum <tobin@ibm.com>
With the changes proposed as part of this PR, a qemu-snp cluster
will be created but no tests will be performed.
GitHub Actions will only run the tests using the workflows that are
part of the **target** branch, instead of the using the ones coming
from the PR. No way to work around this for now.
After this commit is merged, the tests (not the yaml files for the
actions) will be altered in order for the checkout action to help in
this case.
Fixes: #6722
Signed-off-by: Ryan Savino <ryan.savino@amd.com>
With the changes proposed as part of this PR, a qemu-sev cluster will
be created but no tests will be performed.
GitHub Actions will only run the tests using the workflows that are
part of the **target** branch, instead of the using the ones coming
from the PR. No way to work around this for now.
After this commit is merged, the tests (not the yaml files for the
actions) will be altered in order for the checkout action to help in this
case.
Fixes: #6711
Signed-off-by: Ryan Savino <ryan.savino@amd.com>
Now that the infra for running dragonball tests has been enabled, let's
actually make sure to have them running on each PR.
The tests skipped are:
* `k8s-cpu-ns.bats`, as CPU resize doesn't seem to be yet properly
supported on runtime-rs
* https://github.com/kata-containers/kata-containers/issues/6621Fixes: #6605
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
With the changes proposed as part of this PR, an AKS cluster will be
created but no tests will be performed.
The reason we have to do this is because GitHub Actions will only run
the tests using the workflows that are part of the **target** branch,
instead of the using the ones coming from the PR, and we didn't find yet
a way to work this around.
Once this commit is in, we'll actually change the tests themselves (not
the yaml files for the actions), as those will be the ones we want as
the checkout action helps us on this case.
Fixes: #6583
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
The first part of simplifying things to have all our tests using GitHub
actions is moving the k8s tests to this repo, as those will be the first
vict^W targets to be migrated to GitHub actions.
Those tests have been slightly adapted, mainly related to what they load
/ import, so they are more self-contained and do not require us bringing
a lot of scripts from the tests repo here.
A few scripts were also dropped along the way, as we no longer plan to
deploy kubernetes as part of every single run, but rather assume there
will always be k8s running whenever we land to run those tests.
It's important to mention that a few tests were not added here:
* k8s-block-volume:
* k8s-file-volume:
* k8s-volume:
* k8s-ro-volume:
These tests depend on some sort of volume being created on the
kubernetes node where the test will run, and this won't fly as the
tests will run from a GitHub runner, targetting a different machine
where kubernetes will be running.
* https://github.com/kata-containers/kata-containers/issues/6566
* k8s-hugepages: This test depends a whole lot on the host where it
lands and right now we cannot assume anything about that anymore, as
the tests will run from a GitHub runner, targetting a different
machine where kubernetes will be running.
* https://github.com/kata-containers/kata-containers/issues/6567
* k8s-expose-ip: This is simply hanging when running on AKS and has to
be debugged in order to figure out the root cause of that, and then
adapted to also work on AKS.
* https://github.com/kata-containers/kata-containers/issues/6578
Till those issues are solved, we'll keep running a jenkins job with
hose tests to avoid any possible regression.
Last but not least, I've decided to **not** keep the history when
bringing those tests here, otherwise we'd end up polluting a lot the
history of this repo, without any clear benefit on doing so.
Signed-off-by: Fabiano Fidêncio <fabiano.fidencio@intel.com>
