🚀 Release 1.6.0 (2023-11-13)

🗒 NOTES

StackGres 1.6.0 is out! 🎊 🍾

With great pleasure and apologizing for the unexpected long wait, we are proud of this release that brings among other features stability and a much better installation experience thanks to the ability of the operator to reconfigure dynamically without the need of helm, by just modifying the new SGConfig CRD.

Pods resources configuration is quite hard and our approach was not very neat. So we decided to change it by making easy to know the total amount of CPU and memory that will be requested by the Pod using SGInstanceProfile.spec.requests.cpu and SGInstanceProfile.spec.requests.memory as the total where other sidecars will subtract their respective amounts in order to leave the rest for the postgres (patroni) container. Existing resources will still use the previous method to calculate the requests (we provided the flag SGCluster.spec.pods.resources.disableResourcesRequestsSplitFromTotal if you still want to use such method). Limits will still works as expected with SGInstanceProfile.spec.cpu and SGInstanceProfile.spec.memory targeting the patroni container only.

We could not leave the sharded clusters without backups and 2 day operation so we created a bunch of them in order to deal with restarts, resharding and be able to restore from a backup.

And this is yet not over do not wait any more and try out this new shining release of StackGres to find out many other new features, stability improvements and bug fixes!

✨ NEW FEATURES AND CHANGES

• Added PostgreSQL 16.1, 15.5, 14.10, 13.13, 12.17
• Update Patroni to 3.2.0
• Update PgBouncer to 1.21.0
• Support for Kubernetes 1.28
• Support for OpenShift 4.13
• Updated all base images
• Dynamic operator configuration with SGConfig CRD
• Support for sharded backups with SGShardedBackup
• Support for sharded dbops with SGShardedDbOps for restart and resharding operations
• Support for Service Binding
• Support for cluster profile and new calculation for patroni resources
• Improve reconciliation cycle by avoiding rely on custom comparators to detect changes
• Move owner reference decorator as part of the reconciliation logic
• Replaced helm operator with the operator for the operator bundle
• Added skipRange lower bound for operator bundle
• Support only amd64 and arm64 archs for operator bundle
• Avoid check extensions index when extensions did not change
• Improve log output for fluent-bit container
• Make sure JVM and native images set a limit when only requests is set
• Added missing fields to the demo cluster helm chart
• Added dryRun parameter to all create, update and delete endpoints in StackGres REST API
• Added endpoint to create namespaces in StackGres REST API
• Added user management endpoints in StackGres REST API
• Added Grafana dashboards per section with new queries and reduced cardinality of some metrics
• Improvements for majorVersionUpgrade SGDbOps process
• Allow SGShardedCluster to change citus version
• Added sgpostgresconfig(s) as shortNames for SGPostgresConfig
• Added sgpoolingconfig(s) as shortNames for SGPoolingConfig
• Always set latest as the version of resources that can not be upgraded with SGDbOps

Web Console

• Support for SGShardedCluster overrides
• Upgrade VueJS to version 2.7.14

🐛 FIXES

• Major version upgrade fail to show logs and do not complete rollback on error
• create-backup.sh script fail to update message on error
• Add "streaming", "in archive recovery" as MemberState.RUNNING used by patroni since 3.0.4
• Added check for custom resources with too old versions
• PgBouncer configuration not reloaded
• Some prometheus queries are broken
• Container postgres-util is not removed when disabled
• Init container pgbouncer-auth-file is not removed when disableConnectionPooling is set
• Sharded cluster REST API does not load or store scripts for overrides
• Versioned core and contrib extensions are not allowed to change version and unversioned pick random version
• Skip check on resources that do not require a SGDbOps to upgrade if version less or equals to 1.5

Web Console

• Disable editing/cloning/deleting coordinator and shards clusters from a SGShardedCluster
• Improve Distributed Logs location on SGCluster form
• Operator fails to edit a cluster created from a backup (again)
• Move Custom sidecars section (pods) to the existing Sidecars section
• Allow users to revert non-required dropdown selection
• Prevent users from selecting label on timeout selects
• Reuse not found component
• Avoid overwriting syncInstances value when already set
• Prevent setting undefined configurations on sgshardedclusters
• Wrong path on sharded cluster breadcrumbs
• Wrong init of replication mode on sgshardedcluster edition
• Ensure proper storageClass init for coordinator and shards on sgshardecluster edition

🚧 KNOWN ISSUES

• Backups may be restored with inconsistencies when performed with a Postgres instance running on a different architecture (#1539)

🆙 UPGRADE

To upgrade from a previous installation of the StackGres operator's helm chart you will have to upgrade the helm chart release. For more detailed information please refer to our documentation.

To upgrade StackGres operator's (upgrade only works starting from 1.1 version or above) helm chart issue the following commands (replace namespace and release name if you used something different):

helm upgrade -n "stackgres" "stackgres-operator" https://stackgres.io/downloads/stackgres-k8s/stackgres/1.6.0/helm/stackgres-operator.tgz

IMPORTANT: This release is incompatible with previous alpha or beta versions. Upgrading from those versions will require uninstalling completely StackGres including all clusters and StackGres CRDs (those in stackgres.io group) first.

Thank you for all the issues created, ideas, and code contributions by the StackGres Community!

🔀 FULL LIST OF COMMITS