OnGres Inc./StackGres: Release 0.9

Release: 0.9

Released: 2020-08-12

License: AGPL-3.0

Release Assets:

# Release 0.9 ## NOTES Here it comes StackGres 0.9!! :tada: :bottle_with_popping_cork: The most prominent new feature in this release is easy access to postgres logs with an easy to use interface (through the we UI). It is backed by a dedicated postgres instance, with his own special CR called `SGDistributedLogs`. But even if that is the main feature of this release there are many other new features, changes and fixes that we are proud to bring to you. To make this release our entire team did a great job, so we hope you will enjoy it! ## UPGRADE To upgrade from a previous version you will have to re-install StackGres operator. Hot upgrades will be supported in the upcoming `1.0` version. Remember to backup any `SGCluster` data and StackGres CRs. For more detailed information please refer to [our documentation](https://stackgres.io/doc/latest/03-production-installation/02-installation-via-helm/#upgrade-operator). To re-install StackGres issue following commands (replace namespace and release name if you used something different): ``` NAMESPACE=stackgres RELEASE=stackgres-operator helm uninstall -n "$NAMESPACE" "$RELEASE" kubectl delete --ignore-not-found \ customresourcedefinition.apiextensions.k8s.io/sgbackupconfigs.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sgbackups.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sgclusters.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sgdistributedlogs.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sginstanceprofiles.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sgpgconfigs.stackgres.io \ customresourcedefinition.apiextensions.k8s.io/sgpoolconfigs.stackgres.io helm install -n "$NAMESPACE" "$RELEASE" https://stackgres.io/downloads/stackgres-k8s/stackgres/0.9/helm-operator.tgz ``` ## CHANGES * Logs can now be collected and analyzed in the UI or REST API * Patroni 1.6.5 * WAL-G 0.2.15 * Clusters uses Postgres version 11.8 and 12.3 * Clusters uses Envoy version 1.15.0 with new [postgres Envoy network filter](https://www.envoyproxy.io/docs/envoy/v1.15.0/configuration/listeners/network_filters/postgres_proxy_filter) * StackGres pods now run with non-root user (or arbitrary user in OpenShift) * OpenShift 3.11 support added. Also Minishift is added to StackGres integrated test suite. * Custom annotations and labels can now be specified for StackGres cluster pods. * Reorganization of internal pod ports so that 5432 now points to postgres instance. * Added scripts section for cluster initialization to load SQL snippets or small SQL files from ConfigMaps or Secrets * New UI design for the web console * UI and REST API now run in its own pod * UI authentication / authorization based on JWT token and backed by kubernetes RBAC * UI's URLs reload to the same page so they can be used for collaboration (share links to parts of the web console) * UI include now interface to create/edit/delete distributed logs CRDs * Match functionality between CRDs and UI (they are fully equivalent; use either, see the results on either too) * Stats are now shown per Pod and have been improved * Added distributed logs create/edit/view to the UI * Custom Grafana dashboard is used by default when enabling Grafana integration * Timeline is now exposed in backup status * Allow GCP workload identity configuration for Postgres Backups * Update default values for `nonProductionOptions` to be production oriented by default ## FIXES * When deleting a SGBackup CR, the backup in the object storage is not deleted * Automatic backups are removed when the cluster is removed * Manual backups are apparently not working * Connection does not work when SG cluster pod is "Ready" * Creation of cluster for pg 11 with default configuration fail * Fixed retention not honored * Internal error when deleting a distributed logs CR * Cluster resources are not deleted when deleting a cluster from the REST API * Distributed logs REST API return an error relation does not exists * Control Data field is not being stored in the backup CR * Fields are not written in some versions of kubernetes generating cluster update events * Automatic backup job does not enforce a non-root security context * Empty deleted cluster name after deletion * Automatic refreshment is not working * Operator does not update correctly null values * Using fixed version 8.13.4 for prometheus-operator dependency * Using fixed version 5.0.26 for MinIO dependency # KNOWN ISSUES * Kubernetes 1.18 is not supported yet, see #439 * Kubernetes 1.11 requires PodShareProcessNamespace feature gate to be enabled (not tested in any kubernetes environment except for MiniShift 3.11)