feat: optional read-only root filesystem for run pods in the Kubernetes supervisor

Aberta
  • Algo errado?
#5,078 0 comentários 0 reações 0 responsáveis Ver no GitHub

Mantenedores costumam responder em até 1 dia

Ninguém assumiu esta issue ainda.

Avaliação

Dificuldade
3/5
Tempo estimado
1-2 dias
Facilidade para iniciantes
72/100
Tipo de issue
Funcionalidade
Clareza
Claramente especificada
Status de atividade
Ativa
Stack de tecnologia
kubernetes, typescript
Domínio
backend, infrastructure

Direção de pesquisa

Start with apps/supervisor/src/env.ts, workloadManager/kubernetes.ts, and workloadManager/kubernetesPodSpec.ts; trace how runner security context settings and pod volumes are configured. Check the existing unit tests for runnerSecurityContext and the proposed runnerTmpVolume helper. Done means both opt-in environment variables work as described, default behavior stays unchanged, and the related tests pass.

Escrita pelo modelo de indexação a partir do texto da issue.

Descrição

Is your feature request related to a problem? Please describe.

Self-hosters running Trigger.dev on Kubernetes with Pod Security / policy engines (Kyverno, OPA Gatekeeper, CIS benchmarks, some compliance scanners) are often required to run containers with readOnlyRootFilesystem: true. The supervisor's KUBERNETES_RUNNER_SECURITY_CONTEXT (off/baseline/restricted) doesn't set it, and no Pod Security Standards level requires it, so there is currently no way to get it on run pods without patching the supervisor.

Describe the solution you'd like to see

Add two opt-in supervisor env vars:

  • KUBERNETES_RUNNER_READ_ONLY_ROOT_FILESYSTEM (boolean, default false). When true, sets securityContext.readOnlyRootFilesystem: true on the run-controller container. It composes with any KUBERNETES_RUNNER_SECURITY_CONTEXT level, including off.
  • KUBERNETES_RUNNER_TMP_VOLUME_SIZE_LIMIT (default 256Mi, blank = no per-volume cap). When read-only is enabled, an emptyDir is mounted at /tmp so tasks keep a writable os.tmpdir().

The runner runtime itself doesn't write to disk (IPC over socketpair, logs to stdout, OTLP over HTTP), so a single /tmp volume is sufficient. Default behavior is unchanged.

Describe alternate solutions
  • A mutating admission webhook injecting readOnlyRootFilesystem and a /tmp volume. This works but adds cluster-side machinery for something the supervisor can do natively.
  • Maintaining a fork of the supervisor. This is what we do today and what we'd like to stop doing.
  • Making KUBERNETES_RUNNER_SECURITY_CONTEXT=restricted imply the flag. This would break existing users, since restricted doesn't imply a read-only root.
Additional information

We already run this in a private fork (including unit tests for runnerSecurityContext and the new runnerTmpVolume helper). The change touches apps/supervisor/src/env.ts, workloadManager/kubernetes.ts and workloadManager/kubernetesPodSpec.ts, and is roughly 60 lines of source. If this direction is acceptable, we'd be happy to open a PR once vouched, so we can drop our fork.

Linguagem predominante
TypeScript
Estrelas
16.5k
Forks
1.5k
Merge médio
1d 15h
PRs com merge (30d)
17

Preparar o ambiente

Este projeto não oferece contêiner de desenvolvimento, Dockerfile nem guia de contribuição, então a configuração fica por sua conta: comece pelo README e veja nosso guia da primeira contribuição para os passos gerais.

Primeiros passos

  1. Leia a issue inteira e depois o guia de contribuição do projeto.
  2. Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
  3. Faça um fork do repositório e trabalhe em uma branch.
  4. Abra um pull request que referencie o número da issue.

Mais de triggerdotdev/trigger.dev

Todas as issues de triggerdotdev/trigger.dev

Issues semelhantes

Mais issues de TypeScript

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.