WEB_COMPUTE builds stuck at PROVISIONING indefinitely in us-east-1 (April 13, 2026)

[najimovi]

Description

All WEB_COMPUTE builds for our production Amplify app in us-east-1 are getting stuck at the PROVISIONING phase indefinitely with completely empty build logs. The container is never allocated, so no build commands execute.

Details

• App ID: d19igjp5uug58d
• Region: us-east-1
• Platform: WEB_COMPUTE (Next.js SSR)
• 7 consecutive jobs (214–220) all stuck at PROVISIONING since ~16:26 UTC, April 13, 2026
• Staging app (d2tg57oc0x7gw2) in us-west-1 deploys the same codebase successfully in ~7 minutes (just verified)
• No changes to app config, IAM role, buildspec, or code between last successful deploy (Job 213, April 10) and first failure
• Even a RETRY of the last successful job (213) gets stuck at PROVISIONING
• IAM role is valid, all account quotas are healthy (3/25 apps, 1/5 concurrent jobs)

What we've tried

1. RELEASE jobs with and without --commit-id
2. RETRY of last successful job (213)
3. Verified IAM role, policies, and service quotas — all healthy
4. Confirmed staging (us-west-1) deploys successfully with identical code

Suspected cause

StatusGator reports active CodeBuild degradation in Virginia (us-east-1) today, with multiple user reports of "CodeBuild is not working" and "Service down" starting ~17:15 UTC. Since Amplify WEB_COMPUTE uses CodeBuild for container provisioning, this would explain the hang.

Environment

• Build image: amplify:al2023
• Node 22, Yarn 4 (corepack)
• Framework: Next.js 16 (SSR)
• Last successful build: Job 213, April 10, 2026 (~7 min total)

[najimovi]

Update: CodeBuild in us-east-1 recovered as of ~18:53 UTC (April 13, 2026).

• Jobs 214–225 all stuck on PROVISIONING with empty build logs (spanning ~3 hours)
• Job 225 (probe at 18:53 UTC) went straight to RUNNING — first successful provisioning
• Job 226 (full production deploy via GHA workflow) completed successfully: SUCCEED

No configuration, IAM, or code changes were needed on our side — this was purely an AWS CodeBuild infrastructure issue in us-east-1. Confirmed by third-party status trackers (StatusGator, IsDown.app) reporting CodeBuild degradation during the same window.

Leaving this open for visibility in case others hit the same issue today.