Today is my last day working full time on Pinniped (for now). This change removes me from the MAINTAINERS.md and the website.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
These fields were changed as a minor hardening attempt when we switched to Distroless, but I bungled the field names and we never noticed because Kapp doesn't apply API validations.
This change fixes the field names so they act as was originally intended. We should also follow up with a change that validates all of our installation manifest in CI.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
We no longer have a transitive dependency on this older repository, so we don't need the replace directive anymore.
There is a new fork of this that we should move to (https://github.com/golang-jwt/jwt), but we can't easily do that until a couple of our direct dependencies upgrade.
This is a revert of d162cb9adf.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
At a high level, it switches us to a distroless base container image, but that also includes several related bits:
- Add a writable /tmp but make the rest of our filesystems read-only at runtime.
- Condense our main server binaries into a single pinniped-server binary. This saves a bunch of space in
the image due to duplicated library code. The correct behavior is dispatched based on `os.Args[0]`, and
the `pinniped-server` binary is symlinked to `pinniped-concierge` and `pinniped-supervisor`.
- Strip debug symbols from our binaries. These aren't really useful in a distroless image anyway and all the
normal stuff you'd expect to work, such as stack traces, still does.
- Add a separate `pinniped-concierge-kube-cert-agent` binary with "sleep" and "print" functionality instead of
using builtin /bin/sleep and /bin/cat for the kube-cert-agent. This is split from the main server binary
because the loading/init time of the main server binary was too large for the tiny resource footprint we
established in our kube-cert-agent PodSpec. Using a separate binary eliminates this issue and the extra
binary adds only around 1.5MiB of image size.
- Switch the kube-cert-agent code to use a JSON `{"tls.crt": "<b64 cert>", "tls.key": "<b64 key>"}` format.
This is more robust to unexpected input formatting than the old code, which simply concatenated the files
with some extra newlines and split on whitespace.
- Update integration tests that made now-invalid assumptions about the `pinniped-server` image.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
We accidentally missed this in the v0.10.0 release process. The new YAML field here should make it easier to automate this step, which seems like a really good idea.
This may be a temporary fix. It switches the manual auth code prompt to use `promptForValue()` instead of `promptForSecret()`. The `promptForSecret()` function no longer supports cancellation (the v0.9.2 behavior) and the method of cancelling in `promptForValue()` is now based on running the blocking read in a background goroutine, which is allowed to block forever or leak (which is not important for our CLI use case).
This means that the authorization code is now visible in the user's terminal, but this is really not a big deal because of PKCE and the limited lifetime of an auth code.
The main goroutine now correctly waits for the "manual prompt" goroutine to clean up, which now includes printing the extra newline that would normally have been entered by the user in the manual flow.
The text of the manual login prompt is updated to be more concise and less scary (don't use the word "fail").
Signed-off-by: Matt Moyer <moyerm@vmware.com>
- Remove all the "latest" links and replace them with our new shortcode so they point at the latest release in a more explicit way.
This also eliminates one of the sections in our Concierge and Supervisor install guides, since you're always installing a specific version.
- Provide instructions for installing with both kapp (one step) and kubectl (two steps for the Concierge).
- Minor wording changes. Mainly we are now a bit less verbose about reminding people they can choose a different version (once per page instead of in each step).
- When we give an example `kapp deploy` command, don't suggest `--yes` and `--diff-changes`.
Users can still use these but it seems overly verbose for an example command.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This test is asynchronously waiting for the controller to do something, and in some of our test environments it will take a bit longer than we'd previously allowed.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Prior to this fix, this controller did not correctly react to changes to the ClusterIP service. It would still eventually react with a long delay due to our 5 minute resync interval.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This functioned fine, but did not have the intended visual appearance when it came to how the text of the auth code wrapped inside the copy button in the manual flow.
The new styling behaves correctly on at least Chrome, Firefox, and Safari on macOS.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
It turns out that `syscall.Stdin` is of type `int` on Linux and macOS, but not on Windows (it's `syscall.Handle`). This should now be portable and do all the require type casting on every platform.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Our actual CLI code behaved correctly, but this test made some invalid assumptions about the "upstream" IDP we're testing. It assumed that the upstream didn't support `response_mode=form_post`, but Okta does. This means that when we end up on the localhost callback page, there are no URL query parameters.
Adjusting this regex makes the test pass as expected.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
I found that there are some situations with `response_mode=form_post` where Chrome will open additional speculative TCP connections. These connections will be idle so they block server shutdown until the (previously 5s) timeout. Lowering this to 500ms should be safe and makes any added latency at login much less noticeable.
More information about Chrome's TCP-level behavior here: https://bugs.chromium.org/p/chromium/issues/detail?id=116982#c5
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Using the same fake TTY trick we used to test LDAP login, this new subtest runs through the "manual"/"jump box" login flow. It runs the login with a `--skip-listen` flag set, causing the CLI to skip opening the localhost listener. We can then wait for the login URL to be printed, visit it with the browser and log in, and finally simulate "manually" copying the auth code from the browser and entering it into the waiting CLI prompt.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This flag is (for now) meant only to facilitate end-to-end testing, allowing us to force the "manual" login flow. If it ends up being useful we can un-hide it, but this seemed like the safest option to start with.
There is also a corresponding `--oidc-skip-listen` on the `pinniped get kubeconfig` command.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
For some reason our headless Chrome test setup behaves slightly differently on Linux and macOS hosts. On Linux, the emoji characters are not recognized as valid text, so they are URL encoded. This change updates the test to cope with both cases correctly.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This adds a new login flow that allows manually pasting the authorization code instead of receiving a browser-based callback.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This is a more restrictive library interface that more closely matches the use cases of our new form_post login flow.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This allows the prompts to be cancelled, which we need to be able to do in the case where we prompt for a manually-pasted auth code but the automatic callback succeeds.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This is a new pacakge internal/oidc/provider/formposthtml containing a number of static files embedded using the relatively recent Go "//go:embed" functionality introduced in Go 1.16 (https://blog.golang.org/go1.16).
The Javascript and CSS files are minifiied and injected to make a single self-contained HTML response. There is a special Content-Security-Policy helper to calculate hash-based script-src and style-src rules.
This new code is covered by a new integration test that exercises the JS/HTML functionality in a real browser outside of the rest of the Supervisor.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Our Supervisor callback handler now needs to load JS and CSS from the provider endpoint, and this JS needs to make a `fetch()` call across origins (to post the form to the CLI callback). This requires a custom Content-Security-Policy compared to other pages we render.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This test would occasionally flake for me when running locally. This change moves more of the assertions into the "eventually" loop, so they can temporarily fail as long as they converge on the expected values.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This test did not tolerate this connection failing, which can happen for any number of flaky networking-related reasons. This change moves the connection setup into an "eventually" retry loop so it's allowed to fail temporarily as long as it eventually connects.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This CredentialIssuer field is called `spec.impersonationProxy.service.type`, not `spec.impersonationProxy.service.mode`.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This fixes some rare test flakes caused by a data race inherent in the way we use `assert.Eventually()` with extra variables for followup assertions. This function is tricky to use correctly because it runs the passed function in a separate goroutine, and you have no guarantee that any shared variables are in a coherent state when the `assert.Eventually()` call returns. Even if you add manual mutexes, it's tricky to get the semantics right. This has been a recurring pain point and the cause of several test flakes.
This change introduces a new `library.RequireEventually()` that works by internally constructing a per-loop `*require.Assertions` and running everything on a single goroutine (using `wait.PollImmediate()`). This makes it very easy to write eventual assertions.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Before this change, we used the `fosite.DefaultOpenIDConnectClient{}` struct, which implements the `fosite.Client` and `fosite.OpenIDConnectClient` interfaces. For a future change, we also need to implement some additional optional interfaces, so we can no longer use the provided default types. Instead, we now use a custom `clientregistry.Client{}` struct, which implements all the requisite interfaces and can be extended to handle the new functionality (in a future change).
There is also a new `clientregistry.StaticRegistry{}` struct, which implements the `fosite.ClientManager` and looks up our single static client. We could potentially extend this in the future with a registry backed by Kubernetes API, for example.
This should be 100% refactor, with no user-observable change.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
There was nothing to guarantee that _all_ Supervisor pods would be ready to handle this request. We saw a rare test flake where the LDAPIdentityProvider was marked as ready but one of the Supervisor pods didn't have it loaded yet and returned an HTTP 422 error (`Unprocessable Entity: No upstream providers are configured`).
Signed-off-by: Matt Moyer <moyerm@vmware.com>
The `require.Eventually()` function runs the body of the check in a separate goroutine, so it's not safe to use other `require` assertions as we did here. Our `library.RequireEventuallyWithoutError()` function does not spawn a goroutine, so it's safer to use here.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
When a CredentialIssuer is switched from one service type to another (or switched to disabled mode), the `impersonatorconfig` controller will delete the previous Service, if any. Normally one Concierge pod will succeed to delete this initially and any other pods will see a NotFound error.
Before this change, the NotFound would bubble up and cause the strategy to enter a ErrorDuringSetup status until the next reconcile loop. We now handle this case without reporting an error.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
This typo wasn't caught in testing because 1) the Kubernetes API ignores the unknown field and 2) the `type` field defaults to `LoadBalancer` anyway, so things behave as expected.
Even though this doesn't cause any large problems, it's quite confusing.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
The LastUpdateTime is no longer updated on every resync. It only changes if the underlying status has changed, so that it effectively shows when the transition happened.
This change happened in ab750f48aa, but we missed this test. It only fails when it has been more than ten minutes since the CredentialIssuer transitioned into a healthy state, but that can happen in our long-running CI environments.
Signed-off-by: Matt Moyer <moyerm@vmware.com>
Matt MoyerandGitHub
