Engineering Craft › Branching & Releases · also in CI/CD & Deployment, Working in Production

Hotfix

An urgent fix shipped outside the normal release cycle.

Also known as: hotfix branch, emergency fix, urgent fix

A hotfix is an urgent fix for a serious problem in production, shipped outside the normal schedule. Think of a payment bug, a security hole, or an outage, where waiting for the next release costs too much.

The usual flow

git switch main                      # or the release branch / tag that's in production
git pull
git switch -c hotfix/checkout-crash
# make the smallest change that fixes it, plus a test
git push -u origin hotfix/checkout-crash
# fast review, merge, deploy, tag

If the release branch has drifted from main, branch from the production tag, then apply the fix to both so it isn’t lost in the next release.

Principles

• Fix only the problem. No refactors, no extra features. A smaller change is easier to review and less risky.
• Consider a rollback first. If the last release caused it, rolling back is often faster and safer (rollback, mitigate first).
• Still review and test, but streamlined. A second pair of eyes catches mistakes made under pressure.
• Add a regression test so it can’t return (regression tests).
• Deploy carefully and watch: check error rates after release.
• Port it everywhere: merge the fix back into main (and any other maintained branches).
• Tag the release (tags).
• Communicate: tell stakeholders, and follow up with a postmortem if it was an incident.

Cautions

• Hotfixes done in a rush create their own bugs. If the process is too fast, it’s a risk.
• If you have many hotfixes, the real problem is upstream: testing, release process or review.
• A good deployment pipeline makes hotfixes boring, since the same path works at any speed (continuous delivery).