Rollback Procedures

When This Applies

Apply this guidance when:

• Planning rollback strategies before a merge
• Reverting a problematic release
• Handling failed deployments
• Managing emergency hotfixes
• Preparing disaster recovery plans

Rollback Strategy Selection

Decision Tree

Was the issue detected before push to production?
  YES → git reset or revert locally
  NO  → Was data affected?
    NO  → Simple git revert
    YES → Data migration reversal needed
      → Is the migration reversible?
        YES → Run reverse migration + git revert
        NO  → Manual data fix + hotfix

Rollback Methods

Git Revert Process

Simple Revert (single commit)

git checkout main
git revert <commit-hash>
git push origin main

Merge Revert (undo a merge)

# -m 1 means keep the main branch's parent
git checkout main
git revert -m 1 <merge-commit-hash>
git push origin main

After Reverting a Merge

If the original work needs to be re-merged later:

# First, revert the revert to restore the changes
git revert <revert-commit-hash>
# Then apply fixes
# Then merge normally

Hotfix Workflow

For urgent fixes that can't wait for the normal workflow:

1. Create hotfix branch from main: git checkout -b hotfix/NNN-description main
2. Apply the minimal fix
3. Run ALL tests
4. Notify Manager (via queue) about the emergency change
5. Merge to main: git checkout main && git merge --no-ff hotfix/NNN-description
6. Push: git push origin main
7. Also merge to development to keep branches in sync
8. Delete hotfix branch
9. Document in CHANGELOG.md and create a post-incident report

Pre-Merge Rollback Planning

Before every merge to main, prepare:

1. Rollback command — The exact git command to revert
2. Verification steps — How to confirm the rollback worked
3. Data considerations — Whether data changes need reversing
4. Communication plan — Who to notify and how
5. Estimated time — How long a rollback would take

Incident Response

If a production issue is discovered after merge:

1. Assess severity — Is the service down? Is data corrupted? Is it cosmetic?
2. Decide action — Rollback vs hotfix vs wait
3. Execute — Perform the chosen action
4. Verify — Confirm the issue is resolved
5. Communicate — Notify all roles via queue with critical priority
6. Document — Write incident report

Severity Guide