fix: push.sh now targets the real Production AWS account, not the old pre-rebuild instance
push.sh was hardcoded to a single shared ECR registry, instance ID, and IAM credential (musehub-infra) — all of which only ever pointed at the Nonproduction account. Running 'push.sh prod' would have deployed to a stale target that doesn't correspond to the real Production account (672469410277) at all.
- Per-environment config (instance ID, ECR registry, AWS profile) for staging vs. prod, since they are separate AWS accounts - Build once, push the same artifact to each target's own ECR registry under its own credentials - prod authenticates via the operator's own musehub-production SSO profile, matching the no-shared-IAM-user design already documented for Production - Added an auth precheck that fails fast with a clear 'aws sso login' message instead of failing deep into the script - deploy.sh: clarified the manual-run example comment; no behavior change, it already receives ECR_IMAGE as a parameter
Semantic Changes
19 symbols
Files Changed
~2
1059 in snapshot
0 comments
muse hub commit comment sha256:7093b91d5684806df3a8d0f4b520e53716bb19f540822aace6e30c5a05cdf6e4 --body "your comment"
No comments yet. Be the first to start the discussion.