v3.0.0

What's Changed

For AWS (EKS) and Azure (AKS), a refresh has been done of the instance types that are used for many of the services classes. In AWS, this means moving to the r6in family for prod10k and above service classes. In Azure this means moving to v5 instances for all service classes and to the Ebdsv5 family for prod10k and above service classes.

Additionally, the prod5k and prod50k service classes have been split into their own node pools to better differentiate their performance.

This refresh was done to better engineer the instance type choices for performance, as well as move to more modern instance types.

Changelog

• For AKS and EKS add 5k and 50k node groups and update instance types by @danny-wall-sol in #102

Full Changelog: v2.4.0...v3.0.0

v2.4.0

What's Changed

• Updated the worker node auto-upgrade in AKS to 'none' to avoid excessive disruption.

Changelog

• Set node_os_upgrade_channel to None in AKS by @xiaocongji in #103

New Contributors

• @xiaocongji made their first contribution in #103

Full Changelog: v2.3.0...v2.4.0

v2.3.0

What's Changed

Updated the logging for AKS to only log cluster autoscaler logs, the rest is optional.

Changelog

• Only log autoscaler logs in AKS, the rest is optional. by @lumberbaron in #100

Full Changelog: v2.2.0...v2.3.0

v2.2.0

What's Changed

Updated AKS logging configuration to reduce logs a bit, and made configuration of disks for worker nodes more flexible to allow for ephemeral or non-ephemeral options.

Changelog

• Update AKS logging to log fewer things, add flexibility in disk config for system pool by @lumberbaron in #94

Full Changelog: v2.1.0...v2.2.0

v2.1.0

What's Changed

• Removed the 'AllMetrics' entry for AKS diagnostic settings to avoid excess costs (they're expensive)

Changelog

• Remove 'AllMetrics' from AKS diagnostic settings by @lumberbaron in #92

Full Changelog: v2.0.0...v2.1.0

v2.0.0

What's Changed

• Moved the use of node pool modules from the cluster modes to the main project for EKS, AKS, and GKE
• Pod Identity is now the only option for workload identity for EKS
• API access management is now the only option for EKS
• Secondary CIDR ranges for services and pods are now configured directly for GKE

Changelog

• Use manually-created secondary CIDR ranges for services an pods in GKE by @lumberbaron in #83
• Simplify GCP service account names to maximize cluster name by @lumberbaron in #85
• Move EKS node group modules into main terraform project by @lumberbaron in #86
• Move GKE node group modules into main terraform project by @lumberbaron in #87
• Move AKS node pool modules into main terraform project by @lumberbaron in #88

Full Changelog: v1.8.0...v2.0.0

v1.8.0

What's Changed

• Added ability to specify a separate CIDR for the cluster's subnet in the AKS project

Changelog

• feature/split_cidr: Split CIDR for VNET and subnet in network module for AKS by @mcdanielajayi in #82
• Add subnet cidr variable to AKS network module by @lumberbaron in #84

New Contributors

• @mcdanielajayi made their first contribution in #82

Full Changelog: v1.7.0...v1.8.0

v1.7.0

What's Changed

Updated variables in GKE project to validate the cluster name length due to a service account name limitation.

Changelog

• DATAGO-79683: Use EKS 1.31 for testing by @nagsubhrajitt in #80
• chore(DATAGO-87752): add restriction on gke cluster name length by @danny-wall-sol in #81

New Contributors

• @danny-wall-sol made their first contribution in #81

Full Changelog: v1.6.0...v1.7.0

v1.6.0

What's Changed

• Added ability to optionally use Pod Identities in EKS.

Changelog

• Include cluster id in module output for AKS. by @lumberbaron in #66
• Switch to IRSAv2/pod identity by @Luay-Sol in #64
• Add description of use of Pod Identities for EKS in README. by @lumberbaron in #67
• Add policy prefix to pod identity roles. by @lumberbaron in #68

New Contributors

• @Luay-Sol made their first contribution in #64

Full Changelog: v1.5.0...v1.6.0

v1.5.0

What's Changed

• Added flag to EKS project to have node group to set name explicitly instead of a prefix. The prefix option is good if updating node groups via Terraform, but has name length limitations as the suffix is quite long. Prefix option has been kept the default for backwards compatibility, but may be swapped on the next major version release.

Changelog

• Refactor EKS node group name by @nagsubhrajitt in #65

Full Changelog: v1.4.0...v1.5.0