Skip to content

v0.20.1

2026-05-06

  • [x] Prod deploy was failing in CI because the snapshot step called aws rds create-db-cluster-snapshot against purple-cat-prod, but prod is a standalone PostgreSQL RDS instance (purple-cat-vtt-db-prod-1), not an Aurora cluster — the cluster command would never resolve a target. Switched the snapshot step in .github/workflows/ci.yml to the instance-level aws rds create-db-snapshot / wait db-snapshot-available pair, defaulting DB_INSTANCE_ID to purple-cat-vtt-db-prod-1 (overridable via the new RDS_INSTANCE_IDENTIFIER repo var; the legacy RDS_CLUSTER_IDENTIFIER is still honoured as a fallback). Updated apps/web/src/__tests__/ciProdBackup.test.ts to pin the new commands and added a negative assertion against the cluster command to prevent regressions. Companion change in IAM: the CI user (purple-cat-platform-ai-user) needs rds:CreateDBSnapshot / rds:DescribeDBSnapshots / rds:AddTagsToResource / rds:ListTagsForResource on the prod instance and its snapshots; attached as inline policy PurpleCatRdsSnapshotForCi.