Day 68 / 100 · Engineering practices · Incidents and blameless postmortems

Fix the system, not the person.

Blameless postmortems, championed by John Allspaw and others, ask how the system allowed a failure, not who to blame for it.

Restore first

Get customers working again. Analysis comes after.

Build the timeline

What happened, when, from logs and people's accounts.

Find the factors

Several contributing causes, not one culprit.

Own the actions

Specific fixes with owners and dates, tracked to done.

Fix the system, not the person.