Releases: dfds-data/mlflow-infrastructure
Update pipelines to use templates and tofu
-
Refactor Azure Pipelines configuration to streamline MLflow image build and push process
-
Remove unnecessary self-repository declaration from Azure Pipelines configuration
-
Add aws_credentials parameter to tofu-check template in Azure Pipelines
-
Add additional parameters for Docker build context and tagging in Azure Pipelines
-
Update Docker push tag to use mlflow_version.mlflow_tag and add new Terraform plan files
-
Remove redundant ECR push task configuration from Azure Pipelines
-
Update source image name to include version tag for ECR push
-
Refactor Docker image tagging step to use a script for improved clarity
-
Fix ECR push configuration to separate source image name and tag
-
Add extra_commands parameter to tofu-check job for automatic approval
Remove conflicting S3 bucket ACL resource from Terraform configuration
This release implements minor changes to the Terraform files regarding S3 access.
If “Bucket owner enforced” is active for your S3 bucket, ACLs are effectively disabled. Attempting to set them via Terraform (aws_s3_bucket_acl) will fail with AccessControlListNotSupported. See hashicorp/terraform-provider-aws#34079
v4.0.0
BREAKING:
- Changed from KIAM auth to IRSA.
See https://wiki.dfds.cloud/en/teams/devex/operations/guides/kiam-to-irsa-migration.
KIAM is deprecated and will be removed in the future. We are now using IRSA and ServiceAccounts to assume roles in
AWS.
