Koios Automation Platform 2026.11.0 — release notes
First stable release, published 2026-10-31 in the private registry ghcr.io/koiosplatform. Install it
from the Koios Operators catalog
(ghcr.io/koiosplatform/koios-operator-catalog@sha256:6cbdb1ec4d9a64a588d8efd52a2a0504fdaedbc16657cba13de3743498915598)
after configuring registry access, see olm-install.md.
Upgrading from the 2026.10.0 demo
Automation hub cache (Redis → Valkey). 2026.11.0 replaces Redis 7.4 with Valkey 8.1
(BSD-3-Clause). Valkey cannot read the RDB file that Redis 7.4 wrote in the hub cache volume, so
after the upgrade the hub Redis pod fails with Can't handle RDB format version 12. The volume only
holds the Pulp cache; delete it once and the hub operator recreates it:
kubectl -n <namespace> scale deployment <hub>-redis --replicas=0
kubectl -n <namespace> delete pvc <hub>-redis-data
The platform Redis and the controller Redis sidecar keep no data and need no action.
PostgreSQL. The managed PostgreSQL is renamed to <platform>-postgres-15 and the data is
migrated and verified automatically (row counts per table). The previous volume is kept until you
delete it; status.retainedPostgresClaims lists it.
Kubernetes without ReadWriteMany storage
Automation hub requests a ReadWriteMany volume by default. On clusters without such a
StorageClass set spec.hub.file_storage_access_mode: ReadWriteOnce; the platform status reports a
pending hub or EDA volume with its access mode.
Highlights
Koios Automation Platform branding in the UI and in the OpenShift console (OLM).
OLM catalog with
stable-YYYY.MMandstable-YYYY.MM-cluster-scopedchannels.Fields of the AnsibleAutomationPlatform form: external Redis for the gateway,
idle_aap,database.postgres_ssl_mode, exposure, gateway, database and status fields.Same templates on OpenShift and Kubernetes: Koios Valkey and Envoy images run as numeric non-root users.
Koios PostgreSQL image: PostgreSQL 15 on Rocky Linux 9 (same interface and data directory as the previous image; existing data is used as is).