Changes Made

Added environment variables to both "Build AMI stage 1" and "Build AMI stage 2" steps in
.github/workflows/testinfra-ami-build.yml:111,120:

env:
AWS_MAX_ATTEMPTS: 10
AWS_RETRY_MODE: adaptive

What This Does

1. AWS_MAX_ATTEMPTS: 10 - The AWS SDK will retry failed API calls up to 10 times before giving up (default is 3)
2. AWS_RETRY_MODE: adaptive - Uses intelligent retry logic that:
   - Backs off exponentially on throttling errors
   - Adapts retry timing based on AWS response patterns
   - Handles transient network issues (like your "unexpected EOF")
   - Works with rate limiting and temporary connectivity problems

Expected Impact

• Fixes the frequent "unexpected EOF" errors during snapshot tagging by automatically retrying when the AWS API connection drops
• No workflow re-runs needed for transient failures - the build will self-heal
• Minimal time overhead - Only adds delay when retries are actually needed
• Works for both stages - Protects both the surrogate build and the stage 2 build

The frequent "unexpected EOF" was tied many times to a specific interaction with aws that doesn't require a full re-run of workflow