This test could fail when the cluster was under heavy load. This could cause kubectl to emit "Throttling request took [...]" logs that triggered a failure in the test.
The fix is to ignore these innocuous warnings.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
We had this code that printed out pod logs when certain tests failed, but it is a bit cumbersome. We're removing it because we added a CI task that exports all pod logs after every CI run, which accomplishes the same thing and provides us a bunch more data.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This allows setting `$PINNIPED_TEST_CLI` to point at an existing `pinniped` CLI binary instead of having the test build one on-the-fly. This is more efficient when you're running the tests across many clusters as we do in CI.
Building the CLI from scratch in our CI environment takes 1.5-2 minutes, so this change should save nearly that much time on every test job.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
We've seen some test flakes caused by this test. Some small changes:
- Use a 30s timeout for each iteration of the test loop (so each iteration needs to check or fail more quickly).
- Log a bit more during the checks so we can diagnose what's going on.
- Increase the overall timeout from one minute to five minutes
Signed-off-by: Matt Moyer <moyerm@vmware.com>
In the case where we are using middleware (e.g., when the api group is
different) in our kubeclient, these error messages have a "...middleware request
for..." bit in the middle.
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
This is probably a good idea regardless, but it also avoids an infinite recursion from IntegrationEnv() -> assertNoRestartsDuringTest() -> NewKubeclient() -> IntegrationEnv() -> ...
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This test could flake in some rare scenarios. This change adds a bunch of retries, improves the debugging output if the tests fail, and puts all of the subtests in parallel which saves ~10s on my local machine.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This test has occasionally flaked because it only waited for the APIService GET to finish, but did not wait for the controller to successfully update the target object.
The new code should be more patient and allow the controller up to 10s to perform the expected action.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This new capability describes whether a cluster is expected to allow anonymous requests (most do since k8s 1.6.x, but AKS has it disabled).
This commit also contains new capability YAML files for AKS and EKS, mostly to document publicly how we expect our tests to function in those environments.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
At the end of the test, wait for the KubeClusterSigningCertificate
strategy on the CredentialIssuer to go back to being healthy, to avoid
polluting other integration tests which follow this one.
We were previously issuing both client certs and server certs with
both extended key usages included. Split the Issue*() methods into
separate methods for issuing server certs versus client certs so
they can have different extended key usages tailored for each use
case.
Also took the opportunity to clean up the parameters of the Issue*()
methods and New() methods to more closely match how we prefer to call
them. We were always only passing the common name part of the
pkix.Name to New(), so now the New() method just takes the common name
as a string. When making a server cert, we don't need to set the
deprecated common name field, so remove that param. When making a client
cert, we're always making it in the format expected by the Kube API
server, so just accept the username and group as parameters directly.
I'm kinda surprised this is working with our current implementation of the
impersonator, but regardless this seems like a step forward.
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
The impersonator_test.go unit test now starts the impersonation
server and makes real HTTP requests against it using client-go.
It is backed by a fake Kube API server.
The CA IssuePEM() method was missing the argument to allow a slice
of IP addresses to be passed in.
These tests occasionally flake because of a conflict error such as:
```
supervisor_discovery_test.go:105:
Error Trace: supervisor_discovery_test.go:587
supervisor_discovery_test.go:105
Error: Received unexpected error:
Operation cannot be fulfilled on federationdomains.config.supervisor.pinniped.dev "test-oidc-provider-lvjfw": the object has been modified; please apply your changes to the latest version and try again
Test: TestSupervisorOIDCDiscovery
```
These retries should improve the reliability of the tests.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Also make each t.Run use its own namespace to slight reduce the
interdependency between them.
Use t.Cleanup instead of defer in whoami_test.go just to be consistent
with other integration tests.
The same coverage that was supplied by
TestCredentialRequest_OtherwiseValidRequestWithRealTokenShouldFailWhenTheClusterIsNotCapable
is now provided by an assertion at the end of TestImpersonationProxy,
so delete the duplicate test which was failing on GKE because the
impersonation proxy is now active by default on GKE.
When testing that the impersonation proxy port was closed there
is no need to include credentials in the request. At the point when
we want to test that the impersonation proxy port is closed, it is
possible that we cannot perform a TokenCredentialRequest to get a
credential either.
Also add a new assertion that the TokenCredentialRequest stops handing
out credentials on clusters which have no successful strategies.
Signed-off-by: Monis Khan <mok@vmware.com>
To make an impersonation request, first make a TokenCredentialRequest
to get a certificate. That cert will either be issued by the Kube
API server's CA or by a new CA specific to the impersonator. Either
way, you can then make a request to the impersonator and present
that client cert for auth and the impersonator will accept it and
make the impesonation call on your behalf.
The impersonator http handler now borrows some Kube library code
to handle request processing. This will allow us to more closely
mimic the behavior of a real API server, e.g. the client cert
auth will work exactly like the real API server.
Signed-off-by: Monis Khan <mok@vmware.com>
The thing we're waiting for is mostly that DNS is resolving, the ELB is listening, and connections are making it to the proxy.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
All controller unit tests were accidentally using a timeout context
for the informers, instead of a cancel context which stays alive until
each test is completely finished. There is no reason to risk
unpredictable behavior of a timeout being reached during an individual
test, even though with the previous 3 second timeout it could only be
reached on a machine which is running orders of magnitude slower than
usual, since each test usually runs in about 100-300 ms. Unfortunately,
sometimes our CI workers might get that slow.
This sparked a review of other usages of timeout contexts in other
tests, and all of them were increased to a minimum value of 1 minute,
under the rule of thumb that our tests will be more reliable on slow
machines if they "pass fast and fail slow".
This time, don't use the Squid proxy if the cluster supports real external load balancers (as in EKS/GKE/AKS).
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This is more than an automatic merge. It also includes a rewrite of the CredentialIssuer API impersonation proxy fields using the new structure, and updates to the CLI to account for that new API.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
I think this is another aspect of the test flakes we're trying to fix. This matters especially for the "Multiple Pinnipeds" test environment where two copies of the test suite are running concurrently.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
If the test is run immediately after the Concierge is installed, the API server can still have broken discovery data and return an error on the first call.
This commit adds a retry loop to attempt this first kubectl command for up to 60s before declaring failure.
The subsequent tests should be covered by this as well since they are not run in parallel.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
We don't support using the impersonate headers through the impersonation
proxy yet, so this integration test is a negative test which asserts
that we get an error.
- The CA cert will end up in the end user's kubeconfig on their client
machine, so if it changes they would need to fetch the new one and
update their kubeconfig. Therefore, we should avoid changing it as
much as possible.
- Now the controller writes the CA to a different Secret. It writes both
the cert and the key so it can reuse them to create more TLS
certificates in the future.
- For now, it only needs to make more TLS certificates if the old
TLS cert Secret gets deleted or updated to be invalid. This allows
for manual rotation of the TLS certs by simply deleting the Secret.
In the future, we may want to implement some kind of auto rotation.
- For now, rotation of both the CA and TLS certs will also happen if
you manually delete the CA Secret. However, this would cause the end
users to immediately need to get the new CA into their kubeconfig,
so this is not as elegant as a normal rotation flow where you would
have a window of time where you have more than one CA.
Should work on cluster which have:
- load balancers not supported, has squid proxy (e.g. kind)
- load balancers supported, has squid proxy (e.g. EKS)
- load balancers supported, no squid proxy (e.g. GKE)
When testing with a load balancer, call the impersonation proxy through
the load balancer.
Also, added a new library.RequireNeverWithoutError() helper.
Signed-off-by: Margo Crawford <margaretc@vmware.com>
Also update concierge_impersonation_proxy_test.go integration test
to use real TLS when calling the impersonator.
Signed-off-by: Ryan Richard <richardry@vmware.com>
There is a new feature in 1.20 that creates a ConfigMap by default in each namespace: https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG/CHANGELOG-1.20.md#introducing-rootcaconfigmap
This broke this test because it assumed that all the ConfigMaps in the ephemeral test namespace were those created by the test code. The fix is to add a test label and rewrite our assertions to filter with it.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Also:
- Shut down the informer correctly in
concierge_impersonation_proxy_test.go
- Remove the t.Failed() checks which avoid cleaning up after failed
tests. This was inconsistent with how most of the tests work, and
left cruft on clusters when a test failed.
Signed-off-by: Ryan Richard <richardry@vmware.com>
Also:
- Changed base64 encoding of impersonator bearer tokens to use
`base64.StdEncoding` to make it easier for users to manually
create a token using the unix `base64` command
- Test the headers which are and are not passed through to the Kube API
by the impersonator more carefully in the unit tests
- More WIP on concierge_impersonation_proxy_test.go
Signed-off-by: Margo Crawford <margaretc@vmware.com>
This change adds a new virtual aggregated API that can be used by
any user to echo back who they are currently authenticated as. This
has general utility to end users and can be used in tests to
validate if authentication was successful.
Signed-off-by: Monis Khan <mok@vmware.com>
I think we were assuming the name of our Concierge app, and getting lucky
because it was the name we use when testing locally (but not in CI).
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
- Watch a configmap to read the configuration of the impersonation
proxy and reconcile it.
- Implements "auto" mode by querying the API for control plane nodes.
- WIP: does not create a load balancer or proper TLS certificates yet.
Those will come in future commits.
Signed-off-by: Margo Crawford <margaretc@vmware.com>
This is a partial revert of 288d9c999e. For some reason it didn't occur to me
that we could do it this way earlier. Whoops.
This also contains a middleware update: mutation funcs can return an error now
and short-circuit the rest of the request/response flow. The idea here is that
if someone is configuring their kubeclient to use middleware, they are agreeing
to a narrow-er client contract by doing so (e.g., their TokenCredentialRequest's
must have an Spec.Authenticator.APIGroup set).
I also updated some internal/groupsuffix tests to be more realistic.
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
I think the reason we were seeing flakes here is because the kube cert agent
pods had not reached a steady state even though our test assertions passed, so
the test would proceed immediately and run more assertions on top of a weird
state of the kube cert agent pods.
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
This allows us to keep all of our resources in the pinniped category
while not having kubectl return errors for calls such as:
kubectl get pinniped -A
Signed-off-by: Monis Khan <mok@vmware.com>