A failed release is retried without backoff

Ouverte
  • Quelque chose ne va pas ?
#578 0 commentaires 0 réactions 0 personnes assignées Voir sur GitHub

Personne n'a encore pris cette issue.

Évaluation

Difficulté
4/5
Temps estimé
3-5 jours
Accessibilité débutants
54/100
Type d'issue
Bug
Clarté
Plutôt claire
Activité
Active
Stack technique
go
Domaine
backend

Piste de recherche

Start by reading the reconciler's watches for custom resources and release Secrets, then trace how failed attempts update the Irreconcilable condition and write release Secrets. Reproduce with WithFailureRollbacks(false) and a chart containing a Service with port 70000; the issue is resolved when repeated failures use backoff instead of immediately requeuing the resource.

Rédigé par le modèle d'indexation à partir du texte de l'issue.

Description

When an install or upgrade keeps failing, for example because the API server rejects one of the rendered resources, the reconciler retries it right away instead of with controller-runtime's backoff. In our case, it retries about every 1.5 seconds for as long as the chart fails, and logs an error and writes to the API server each time.

Two things requeue the custom resource after every failed attempt, and each of them alone keeps the loop going:

  1. The status update. Every failed attempt writes the status of the custom resource, and the reconciler watches the custom resource for any change, including a status change. The status changes even when the error stays the same, because the reconciler sets Irreconcilable to False before each attempt and back to True after it fails, which gives the condition a new lastTransitionTime.
  2. The release Secrets. The custom resource owns its release Secrets, and the reconciler watches them, so the Secret that Helm writes for each failed revision requeues the resource.

This is the problem of #224. Its fixes, #225 and #226, were closed as stale. It also happens with WithFailureRollbacks(false), without rollbacks.

A possible fix is to ignore updates of the custom resource that change only its status, as the reconciler already does for dependent resources, which also covers errors whose message changes between attempts, and to keep the release Secret writes from requeuing the resource.

Steps to reproduce

Reconcile a custom resource with WithFailureRollbacks(false) and a chart that renders a resource the API server rejects, for example a Service with port 70000.

Langage dominant
Go
Étoiles
59
Forks
54
Aucune PR mergée en 30 j

Préparer son environnement

  • Fournit un Dockerfile ou un fichier Docker Compose
  • Aucun modèle de pull request
  • Aucun guide de contribution

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de operator-framework/helm-operator-plugins

Toutes les issues de operator-framework/helm-operator-plugins

Issues similaires

Plus d'issues Go

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.