takescoop/kubectl-exec-forward

A kubectl plugin to run exec hooks exposed by a Kubernetes pod around a port-forward action.

kubectl-exec-forward's Issues

Add Go build/dependency caching

The test workflow is currently pretty slow. go test takes ~2m. There is one slow integration test (TestRunForwardCommand) but the majority of the time is going to downloading dependencies and building packages. Caching this data across builds should significantly reduce test run times.

Run unit and integration test separately

Integration tests depend on Kubernetes (minikube) and are therefore very slow to start. Rather than just allow integration tests to be skipped, we should run unit and integration tests separately, potentially forgoing the inclusion of integration tests in coverage metrics.

This will allow unit tests to run quickly and fail for basic issues, including un-buildable code. Integration tests can either run after or in parallel.

In addition to go test -short, we have the option to use build tags for tests:

https://mickey.dev/posts/go-build-tags-testing/

User agent not set properly on all Kubernetes requests

The calls made by the factory builder appear to be the culprit. We should either ensure that the user agent is set properly for these requests, find a way to make generic object requests using the basic client-go clientset (might not be possible), or find a way to use the dynamicResourceClient

`--context` flag is not used when finding an attachable pod

When calling the plugin with --context, the context setting does not propagate when attempting to find an attachable pod. Given a current context set via kubeconfig of a, and --context b, the resulting behavior is:

• Pod is found using context a
• Port forwarding connection is attempted with context b

When the pod is found using a higher level object, e.g. service/my-svc, both contexts may have the same objects. In this case, an attachable pod will be found in a, but attempts to connect to the named pod in b will likely fail since higher order controllers (Deployment/Replica Set) generate randomized pod names.

Error: error upgrading connection: pods "my-app-abc123" not found

For now, a workaround is to set the context in the kubeconfig, by using kubectl config, kubectx, etc.

All commands (especially hooks) should print logs before they start

When there are pre-connect and post-connect commands that run, the user does not get any indication of what things are running and when. Before each command runs, the program should print relevant log information so the user understands what's running. Child process outputs should generally not be printed, but it's important for users to know what was passed to exec, in case:

• It's slow
• It does something unexpected/malicious

Example output might be:

Running pre-connect hook (id=foo): echo foo bar baz

Rich text rendering could help this display clearly, e.g. bold/colored text. Or quotes/punctuation.

Close forwarding connection when command completes

When the main command completes, the forwarding connection should be closed by default. An ideal command (e.g. psql) should block until work is complete and then exit 0.

For commands that perform a side effect and then immediately exit 0, either the config or the user (CLI) could be allowed to specify that the connection should persist after the command.

Dependency Dashboard

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Rate-Limited

These updates are currently rate-limited. Click on a checkbox below to force their creation now.

• fix(deps): update module github.com/tidwall/gjson to v1.17.1
• fix(deps): update module github.com/ttacon/chalk to v0.1
• chore(deps): update actions/checkout action to v4
• chore(deps): update actions/setup-go action to v5
• chore(deps): update codecov/codecov-action action to v4
• chore(deps): update golangci/golangci-lint-action action to v4
• chore(deps): update goreleaser/goreleaser-action action to v5
• chore(deps): update tibdex/github-app-token action to v2
• 🔐 Create all rate-limited PRs at once 🔐

Open

These updates have all been created already. Click a checkbox below to force a retry/rebase of any.

• fix(deps): update module github.com/stretchr/testify to v1.8.4
• fix(deps): update kubernetes packages to v0.29.2 (k8s.io/api, k8s.io/apimachinery, k8s.io/cli-runtime, k8s.io/client-go, k8s.io/kubectl)
• fix(deps): update module github.com/spf13/cobra to v1.8.0
• Click on this checkbox to rebase all open PRs at once

Detected dependencies

github-actions

.github/workflows/approve-markdown.yml

• actions/checkout v3
• hmarr/auto-approve-action v2

.github/workflows/lint.yml

• actions/checkout v3
• actions/setup-go v3
• golangci/golangci-lint-action v3

.github/workflows/release.yml

• actions/checkout v3
• actions/setup-go v3
• tibdex/github-app-token v1
• goreleaser/goreleaser-action v3

.github/workflows/test.yml

• actions/checkout v3
• actions/setup-go v3
• codecov/codecov-action v3

gomod

go.mod

• go 1.19
• github.com/howeyc/fsnotify v0.9.0
• github.com/pborman/ansi v1.0.0
• github.com/phayes/freeport v0.0.0-20220201140144-74d24b5ae9f5@74d24b5ae9f5
• github.com/spf13/cobra v1.5.0
• github.com/stretchr/testify v1.8.0
• github.com/tidwall/gjson v1.14.3
• github.com/ttacon/chalk v0.0.0-20160626202418-22c06c80ed31@22c06c80ed31
• k8s.io/api v0.25.2
• k8s.io/apimachinery v0.25.2
• k8s.io/cli-runtime v0.25.2
• k8s.io/client-go v0.25.2
• k8s.io/kubectl v0.25.2

---

• Check this box to trigger a request for Renovate to run again on this repository

Command errors should print actual command rather than template

When a command errors, the configuration is printed:

kubectl-exec-forward/internal/command/command.go

Lines 101 to 103 in 981d226

	if err := cmd.Run(); err != nil {
	fmt.Fprintf(streams.ErrOut, "Error running command: %s\n", c)
	fmt.Fprintf(streams.ErrOut, "%s\n", berr)

What the user wants is to see the actual command that ran and failed, not the command template defined by the administrator.