Skip to main content

Troubleshoot Common Errors

First aid: catladder project doctor​

catladder project setup provisions infrastructure for the config at the time it runs. Config edits that change required infrastructure (e.g. adding cloudSql to a component implies the roles/cloudsql.admin IAM role) drift silently until a ci job hits a 403.

catladder project doctor              # whole project
catladder project doctor <component>  # narrow the per-component checks

The doctor is read-only and compares the state the current config implies against what is actually provisioned:

  • catladder store entries (gcloud project numbers) for all cloud run projects
  • gitlab setup outputs: semantic-release token (incl. expiry), GL_TOKEN variable, catladder topic, registry deploy token
  • per component/env: deploy service account exists and carries all config-implied IAM roles; required gcloud services enabled; artifacts registry present
  • github: every secret/variable the generated workflows reference exists in its github environment (names only — github secrets are write-only)

Each problem is reported with the command that heals it (usually catladder project setup <component> or catladder project secrets-sync-github).

Error: UPGRADE FAILED: another operation (install/upgrade/rollback) is in progress​

Might happen if you cancel the deploy job.

  • helm list --namespace <your namespace> --all
  • Check if there is a "pending upgrade" and check the revision number where status is deployed
  • then helm rollback --namespace <your namespace> <name> [<deployed-revision-number>] to roll back to the previous version (the revision number is only required if the deployed revision is not the second last version)
  • trigger the update again

Build fails due to node_modules content​

The node_modules is cached between all pipeline jobs, which means more packages may be present than what is in the lockfile. Yarn should handle this well (pnpm projects cache only the store, not node_modules).

But if you suspect the build fails due to the content of node_modules, clear the CI cache and retry the job(s) — on GitLab clear the cache manually, on GitHub delete the entries under Actions → Caches.

Expired GitLab token​

  1. create new personal access token with api scope https://<your-gitlab-host>/-/user_settings/personal_access_tokens
  2. change it in ~/.config/catladder/preferences.yml

Please add more!