Upgrading

How to upgrade the relay safely.

Standard upgrade

helm upgrade tekton-events-relay \
  oci://ghcr.io/fabioluciano/charts/tekton-events-relay \
  --namespace tekton-events-relay \
  --reuse-values

Pin a chart version with --version x.y.z in production rather than tracking latest. Release notes for every version: Releases.

Before you upgrade

1. Read the release notes for breaking changes to config keys, annotations or defaults.
2. Validate your config against the new version without deploying:

   docker run --rm -v $PWD/config.yaml:/config.yaml \
     ghcr.io/fabioluciano/tekton-events-relay:<new-version> \
     --validate --config /config.yaml

3. Check the Compatibility matrix if you’re also upgrading Kubernetes or Tekton.

What happens to in-flight events

• The relay shuts down gracefully (shutdown_timeout_sec, default 30s) — queued events are processed before exit.
• Tekton retransmits undelivered CloudEvents, and the deduper prevents double-delivery on the new pod — if you use a shared state backend. With the in-memory store, a restart loses dedup state; mode: upsert comments remain safe either way.
• Events that failed permanently before the upgrade stay in the DLQ and can be replayed after.

Rolling back

helm rollback tekton-events-relay --namespace tekton-events-relay

Config is forward-compatible within a minor series; downgrading across minors may reject newer config keys — validation will name the offending key.