What autobase-tech/autobase shipped
Written by FoxPlug from the repository's own releases; not affiliated with Autobase. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/autobase-tech/autobase; Autobase did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.
Get a weekly update like this for your product, free
Week of November 2, 2026
What shipped
- Released version 2.12.0, which adds built-in monitoring dashboards and is recommended for all users to upgrade. Release
- Added PgDog as an opt-in connection pooler and load balancer that can replace PgBouncer and HAProxy, with PgBouncer still the default. Pull request #1781
- Added support for bootstrapping a Patroni standby cluster from a pgBackRest repository using
create_replica_methods, skipping stanza-create. Pull request #1779 - Added stable storage IDs so backup bucket and Azure account names are generated deterministically and stay compatible across providers. Pull request #1770
- Fixed VIP assignment for Patroni mode with TLS enabled by setting the Patroni REST API endpoint and TLS settings. Pull request #1782
- Changed vip-manager to use TLS when talking to Consul and the Patroni REST API. Pull request #1775
- Fixed Consul TLS client variables so they are exported only when TLS is enabled, which stops consul and patronictl from failing on plain HTTP setups. Pull request #1774
- Fixed the Consul HTTP API check condition to work on ansible-core 2.19. Pull request #1773
- Moved privilege grants to run after extension creation so privileges on extension-created objects can be granted on fresh deployments. Pull request #1777
- Fixed check mode so role membership grants are skipped when users are not actually created. Pull request #1778
Why it matters
This week brings a new release, 2.12.0, with built-in monitoring dashboards, alongside several fixes that affect TLS and Consul setups where vip-manager, consul and patronictl previously failed to connect. Standby clusters can now be bootstrapped from pgBackRest, privileges on extension objects can be granted on fresh deployments, and PgDog is available as an opt-in pooler for those who want one process instead of two.
Changelog entry
- Release 2.12.0 with built-in monitoring dashboards. Release
- Add PgDog as an opt-in connection pooler and load balancer. Pull request #1781
- Bootstrap a standby cluster from pgBackRest using
create_replica_methodsand skip stanza-create. Pull request #1779 - Add stable IDs for backup storage names across cloud provisioning, pgBackRest and WAL-G. Pull request #1770
- Fix vip-manager to use the Patroni REST API endpoint and TLS settings. Pull request #1782
- Use TLS towards Consul and the Patroni REST API in vip-manager. Pull request #1775
- Export Consul TLS client variables only when TLS is enabled. Pull request #1774
- Fix the Consul HTTP API check condition on ansible-core 2.19. Pull request #1773
- Grant privileges after creating extensions. Pull request #1777
- Skip role membership grants in check mode. Pull request #1778
- Run
pg_upgradetests on Rocky Linux and AlmaLinux again. Pull request #1780
Autobase 2.12.0 is out with built-in monitoring dashboards. This week also adds PgDog as an opt-in pooler, pgBackRest standby bootstrap, stable backup storage IDs, and fixes for vip-manager and Consul TLS setups.
Autobase 2.12.0 is released with built-in monitoring dashboards, and all users are recommended to upgrade. This week also adds PgDog as an opt-in connection pooler and load balancer, support for bootstrapping a Patroni standby cluster from pgBackRest, and stable storage IDs for deterministic backup bucket and Azure account names. Fixes cover vip-manager using TLS towards Consul and the Patroni REST API, VIP assignment in Patroni mode with TLS, Consul TLS client variables only exported when TLS is enabled, the Consul HTTP API check on ansible-core 2.19, privilege grants after extension creation, check mode role membership, and pg_upgrade tests running again on Rocky Linux and AlmaLinux.
Week of October 26, 2026
What shipped
- Pgbackrest restores now preserve WAL archiving by honoring
disable_archive_commandduring recovery and letting Patroni start targeted PITR replicas, witharchive_commandpreservation made the default. Pull request #1762 - PostgreSQL CLI access now uses dedicated password files via PGPASSFILE instead of PGPASSWORD, keeping passwords out of sudo logs across deployment, maintenance and upgrades. Pull request #1761
- Cluster connection output for PostgreSQL and PgBouncer is now unified, including multi-host URIs with
target_session_attrs, and fixes for DigitalOcean aliases, Consul endpoints, load balancer addresses and public IP reporting. Pull request #1763 - Users can be excluded from the secrets export with secret: false. Pull request #1764
- Extension pre-checks now run on all
postgres_clusternodes, so required libraries are added toshared_preload_librariesbeyond the first master. Pull request #1759 - Deployment now fails when
pg_ctlreload fails on the master, while replicas keep best-effort reloads. Pull request #1760 - The resize filesystem play is now skipped when state is absent comes from the inventory or
group_varsrather than only extra vars. Pull request #1757 - EC2 instances now require IMDSv2 with
http_tokensrequired andhttp_endpointenabled, gated by the newaws_ec2_imds_v2_requiredvariable. Pull request #1756 - SSH access checks now use each host's Ansible connection settings, including
ansible_ssh_common_args, so servers reachable only through a jump host are supported. Pull request #1753 - The
pre_checkspassword-recovery tasks are no longer skipped under --check. Pull request #1755
Why it matters
This week tightens backup and restore behavior, password handling and cluster connection reporting, and extends configuration to more hosts. Users get fewer silent failures and more servers that work out of the box, including jump-host setups and stricter EC2 metadata requirements.
Changelog entry
- Pgbackrest: preserve WAL archiving and prevent replica promotion during restore, honoring
disable_archive_commandand defaulting toarchive_commandpreservation Pull request #1762 - Replace PGPASSWORD with secure password files via PGPASSFILE Pull request #1761
- Unify and correct cluster connection info for PostgreSQL and PgBouncer Pull request #1763
- Exclude a user from the secrets export with secret: false Pull request #1764
- Run extension pre-checks on all
postgres_clusternodes Pull request #1759 - Fail on
pg_hbareload errors on master Pull request #1760 - Skip resize filesystem play when state is absent Pull request #1757
- AWS: require IMDSv2 on EC2 instances Pull request #1756
- Add support for SSH jump host Pull request #1753
- Fix
pre_checkspassword-recovery tasks skipped under --check Pull request #1755
This week: pgbackrest restores preserve WAL archiving, PostgreSQL passwords move to PGPASSFILE, connection output is unified, secret: false excludes users from the secrets export, and extension pre-checks run on all cluster nodes.
This week's changes: pgbackrest restores preserve WAL archiving and use standby mode for latest replica restores, with archive_command preservation now the default. PostgreSQL CLI passwords move from PGPASSWORD to dedicated files via PGPASSFILE, keeping them out of sudo logs. PostgreSQL and PgBouncer connection output is unified, including multi-host URIs. Secret: false excludes a user from the secrets export, extension pre-checks run on all postgres_cluster nodes, and SSH checks honor jump hosts.
Week of October 19, 2026
What shipped
- Community Edition and Enterprise Edition of Autobase Console are being brought together into a single codebase in Autobase Platform. Pull request #1751
- Removal of unreachable cluster nodes now resolves target nodes through cluster membership when host facts are unavailable and removes DCS membership through nodes that can be reached. Pull request #1747
- PostgreSQL upgrades now capture the source cluster's locale provider, provider locale and ICU rules and pass compatible options to initdb for the new cluster. Pull request #1748
- New optional variables choose the initdb locale provider: libc, icu or builtin, with an ICU locale variable required for icu. Pull request #1746
- In value-only mode, the connection info secret is now merged rather than replaced, so only the connection key is rewritten. Pull request #1745
- A template crash in passwords.yml is fixed by guarding
delegate_toagainst an empty group. Pull request #1750
Why it matters
This week mixes consolidation of the Console editions with fixes for node removal and PostgreSQL upgrades. For people running clusters, upgrades keep their locale settings and unreachable nodes can be cleaned up more reliably. The new locale provider options give more control over how new clusters are initialized.
Changelog entry
- Unify Community and Enterprise editions of Autobase Console into a single codebase in Autobase Platform. Pull request #1751
- Fix removal of unreachable cluster nodes by resolving nodes through cluster membership and removing DCS membership through available nodes. Pull request #1747
- Preserve locale provider, provider locale and ICU rules during PostgreSQL upgrades by passing them to initdb. Pull request #1748
- Add optional initdb locale provider variables for libc, icu and builtin, with an ICU locale option required for icu. Pull request #1746
- Merge connection info into its secret in value-only mode, rewriting only the connection key. Pull request #1745
- Guard
delegate_toin passwords.yml against an empty group to avoid a template crash. Pull request #1750
This week: Console Community and Enterprise editions unified in one codebase in Autobase Platform. Also: safer removal of unreachable cluster nodes, locale settings preserved across PostgreSQL upgrades, new initdb locale provider options, and a merge-mode fix for connection secre
This week in Autobase: the Community and Enterprise editions of Autobase Console are being unified into a single codebase in Autobase Platform, simplifying maintenance and moving feature work into one shared implementation. Fixes and features also landed: removal of unreachable cluster nodes now resolves targets through cluster membership and removes DCS membership through reachable nodes, PostgreSQL upgrades preserve the source cluster's locale provider and ICU rules, new optional variables let you choose the initdb locale provider (libc, icu or builtin), value-only mode now merges rather than replaces the connection info secret, and a delegate_to template crash against an empty group is fixed.
Week of October 12, 2026
What shipped
- Cluster servers can now be distributed across multiple availability zones, with AWS as the first supported provider and a default zone count of three. Pull request #1732
- GCP deployments now spread servers across three availability zones by default when a region is given for the server location. Pull request #1738
- Azure VMs can now be placed across availability zones based on SKU support and subscription limits, keeping existing placements, load balancer IPs and backend addresses intact. Pull request #1740
- Hetzner clusters now get a spread placement group per cluster, assigning VMs to different physical hosts within the chosen location. Pull request #1742
- PostgreSQL user entries accept an optional secret parameter so a user can be exported to that named secret instead of the default path. Pull request #1739
- The README has a One-Click Deploy section with a deploy button placed between Getting Started and Compatibility. Pull request #1741
Why it matters
This week extends availability zone handling to AWS, GCP and Azure and adds host spreading on Hetzner, so clusters can be laid out to survive individual host or zone failures. PostgreSQL user secrets can now be routed to specific secrets, and the README adds a faster way to try a deployment.
Changelog entry
- Add multi-AZ server placement with AWS as the first supported provider and
server_zones_countdefaulting to 3. Pull request #1732 - Distribute GCP servers across 3 availability zones by default when
server_locationis a region, configurable viaserver_zones_countandserver_zones. Pull request #1738 - Add multi-AZ support for Azure, placing VMs by SKU and subscription support while preserving existing placement and load balancer addresses. Pull request #1740
- Create a spread placement group per Hetzner cluster and assign its VMs to different physical hosts in the location. Pull request #1742
- Add an optional secret parameter to
postgresql_usersentries for per-user secrets export. Pull request #1739 - Add a One-Click Deploy section with a deploy button to the README. Pull request #1741
Autobase now supports multi-AZ server placement on AWS, GCP and Azure, plus Hetzner spread placement groups for host-level resilience. PostgreSQL users can also be exported to a named secret.
This week's Autobase changes focus on where servers run and how secrets are handled. Multi-AZ support now covers AWS, GCP and Azure, letting clusters spread servers across availability zones with configurable zone counts. Hetzner clusters get a spread placement group so VMs land on different physical hosts. PostgreSQL users accept an optional secret parameter to export a user to a named secret instead of the default path. The README also has a new One-Click Deploy section with a deploy button.
Week of September 21, 2026
What shipped
- Azure VM pricing updated with Dsv6 and Esv6 series including larger 128 and 192 vCPU options, while B-series instances removed due to availability issues. Pull request #1718
- Console now automatically selects provider-specific default cloud regions when creating new clusters instead of requiring manual selection. Pull request #1719
- AWS EC2 instances can now be created in non-default subnets using the new
vpc_subnet_idoption. Pull request #1715 - AWS EC2 volumes now support
volume_nameandsystem_volume_namevariables for better identification. Pull request #1716 - Fixed duplicate host entries in in-memory inventory generation during resizefs operations. Pull request #1717
Why it matters
Azure pricing now reflects current VM availability and larger instance options. AWS deployments gain flexibility for subnet selection and volume naming. Cluster creation becomes faster with automatic region defaults.
Changelog entry
- Azure: Updated VM pricing with Dsv6 and Esv6 series (up to 192 vCPU), removed B-series instances Pull request #1718
- Console: Automatic default region selection for new clusters based on provider preferences Pull request #1719
- AWS: Added
vpc_subnet_idoption for EC2 instance creation in non-default subnets Pull request #1715 - AWS: Added
volume_nameandsystem_volume_namevariables for EC2 volumes Pull request #1716 - Fixed: Duplicate hosts in in-memory inventory during resizefs operations Pull request #1717
This week: Azure pricing refresh with Dsv6/Esv6 series, automatic cloud region selection for new clusters, AWS subnet and volume naming improvements, and inventory deduplication fixes.
This week's updates improve cloud infrastructure flexibility. Azure VM pricing now includes the latest Dsv6 and Esv6 series with larger instance sizes while removing unavailable B-series options. The console automatically selects optimal cloud regions when creating clusters. AWS deployments gain the ability to specify subnets and name volumes. Fixed duplicate host entries in inventory generation.
Week of September 14, 2026
What shipped
- Autobase 2.11.0 released, marking a milestone for full PostgreSQL lifecycle management. Release
- PostgreSQL users role now supports configurable password generation with
postgresql_users_password_generationset to disabled by default,on_createfor new users, or always for rotation. Pull request #1712 - External secrets provider support added through a new vitabaks.autobase.secrets role for exporting and storing secrets in AWS Secrets Manager. Pull request #1711
- Per-replica cloud load balancer controls added with independent settings across all supported cloud providers. Pull request #1695
- AWS support extended to attach multiple Security Groups to EC2 instances and Spot instances via
aws_security_group_idsvariable. Pull request #1707 - EBS encryption now enabled by default with
aws_ebs_encryptedand optionalaws_ebs_kms_key_idfor KMS key support. Pull request #1706 - Post-creation resource tagging removed for Azure, GCP, DigitalOcean, and Hetzner Cloud by moving ownership tags to provider module creation. Pull request #1701
- AWS post-creation EC2 tagging tasks removed as amazon.aws.ec2_instance now applies ownership tags atomically during creation. Pull request #1700
- PostgreSQL users role fixed by adding missing
login_dbparameter to the Grant roles to users task. Pull request #1705 - Firewall hostvars lookup corrected to use
inventory_hostnameinstead ofbind_addressfor consistency across the collection. Pull request #1710
Why it matters
Version 2.11 represents a major step forward for managing PostgreSQL clusters end-to-end. This week's updates add critical security features like external secrets integration and EBS encryption, improve password management flexibility, and fix configuration issues that affected role assignment and firewall setup. Cloud providers also see efficiency gains from eliminating redundant post-creation tagging.
Changelog entry
- PostgreSQL users password generation now configurable via
postgresql_users_password_generation(disabled by default,on_createfor new users, always for rotation) Pull request #1712 - External secrets provider support added for AWS Secrets Manager through new vitabaks.autobase.secrets role Pull request #1711
- AWS: Multiple Security Groups support via
aws_security_group_idsvariable combining user-defined and role-managed groups Pull request #1707 - AWS: EBS encryption enabled by default with
aws_ebs_encryptedand optionalaws_ebs_kms_key_idfor KMS support Pull request #1706 - Per-replica cloud load balancer controls added:
cloud_load_balancer_replica,cloud_load_balancer_replica_sync, andcloud_load_balancer_replica_asyncPull request #1695 - Post-creation tagging removed for Azure, GCP, DigitalOcean, and Hetzner Cloud; ownership tags now passed at resource creation Pull request #1701
- AWS post-creation EC2 tagging removed; ownership tags applied atomically by amazon.aws.ec2_instance Pull request #1700
- PostgreSQL users: Added missing
login_dbparameter to role grant task Pull request #1705 - Firewall configuration: Fixed hostvars lookup to use
inventory_hostnameinstead ofbind_addressPull request #1710 - PgBackRest: Added diff: false to prevent sensitive data logging in diff mode Pull request #1709
- PgBackRest: Added
no_logto prevent sensitive configuration details logging Pull request #1708 - Upgrade: Fixed doubled log path in
recovery_targetLSN check during blue-green major-version upgrade Pull request #1694
Autobase 2.11 ships with AWS Secrets Manager integration, configurable PostgreSQL password generation, EBS encryption by default, and multi-security group support. Plus fixes for replica load balancing controls.
Autobase 2.11 is here. This release brings external secrets management through AWS Secrets Manager, flexible password generation for PostgreSQL users, EBS encryption enabled by default, and independent replica load balancer controls across all cloud providers. We've also eliminated redundant post-creation tagging and fixed critical issues in role assignment and firewall configuration. Full PostgreSQL lifecycle management just got stronger.