On August 17, GitHub went dark for seven hours and forty seven minutes. The most useful sentence in the postmortem it published three days later was not about servers. It was about the rescue attempt, and it is the best short lesson on how to stay calm under pressure as a leader I have read all year.

What actually happened

GitHub is the Microsoft-owned platform where an enormous share of the world’s software gets written, reviewed, and shipped. On August 17 a component in its Central US data center could not keep up with a new traffic peak and started throwing errors. Logins, Actions, pull requests, issues, the API, Copilot, all of it degraded or went down. That part is ordinary. Every system fails eventually.

Here is the part that is not, in GitHub’s own words: “Errors in those services triggered a client-side retry loop that increased traffic during recovery. We had to mitigate that behavior before we could safely restore traffic.”

Read that twice. Everything that got an error tried again. Then it tried again. One internal service that normally handles around 9,000 requests a second was taking closer to 100,000. The system was drowning, and the response to the drowning was more water. GitHub could not restore service until it first got everybody to stop. Not stop sabotaging. Stop helping.

Now put people in the story

I have walked into this room more times than I can count, and there was never a server in it.

Something breaks on Tuesday. By Wednesday there is a standing 8 a.m. sync and a 5 p.m. sync. Three new people are on the thread because they are “close to it.” An executive who heard about it Thursday wants a daily update, which means somebody has to stop fixing the problem in order to describe the problem. Every one of those moves was made by a person trying to help. The fix arrives slower, because the two people who could actually solve it now spend half their day being helped.

That is a retry storm. It just has a calendar invite instead of an HTTP status code.

The two words that separate a leader from a retry loop

Here is the machine version of the mistake. Trigger. Error. Try again. Nothing in between the input and the output. No evaluation, no budget, no decision. Just reflex at the speed of a processor.

Most people run the same loop. Trigger, emotion, action. Something breaks, you feel the drop in your stomach, and your hands are already typing. The meeting gets scheduled before anyone has asked whether a meeting helps.

Leaders run a different sequence, and it is only two words longer.

TRIGGER to EMOTION to THINK/DECIDE to ACT.

That is the first element of the process I teach, and it is first for a reason. Emotions drive our actions, and our actions drive our results. You do not get to skip the emotion. What you get is a choice about whether it goes straight into your hands or passes through your head on the way.

But here is the catch, and it is the part most leadership advice skips. You cannot think and decide about an emotion you have not named. An unacknowledged emotion does not go away. It moves to your unconscious mind and picks your actions for you, and you find out what it picked after the fact. A pile-on is exactly that. It is anxiety, unnamed, scheduling meetings.

How to stay calm under pressure as a leader

I learned the cost of an unnamed emotion at jump school. My first free-fall jump nearly ended my SEAL career, and not for lack of skill. I was afraid, I refused to admit it, and that unacknowledged fear produced inaction at 12,000 feet, where inaction gets people killed. I was told one more jump like that would be my last.

What fixed it was not more jumping. It was twenty brutally honest minutes on the ground, admitting to myself that I was afraid. That was the whole intervention. Once the fear had a name, it stopped running the jump, and I never had a bad one again.

Which brings us to the actual practice, and it is almost embarrassingly simple. Four times a day, in real time, stop and name the emotion you are in. Do not judge it. You are allowed to feel how you feel. Just name it and write it down, one line, with what triggered it. Naming it moves it from the unconscious mind, where it wreaks havoc, to the conscious mind, where you can think and decide.

Four notes a day feels too small to matter. Small is the point. In times of chaos and stress you go to what you know, and the log is how naming what you feel becomes what you know. Do it for a month and the half second shows up on the day everything breaks, because your head has taken that rep a hundred times on days nothing was breaking.

Apply it this week

1. Start the log today. Four times a day, one line, in the moment: the emotion and what triggered it. No judging. Friday, read the week back and look for the pattern. That pattern is what has been picking your actions for you.
2. Add nothing for ten minutes. The first ten minutes after something breaks are a no-build zone. No new meeting, no new thread, no new person. Ten minutes will not cost you the fix. The pile-on might.
3. Ask what is already in flight. Say it out loud, in those words. You will find two or three efforts nobody knew were running in parallel. Cut the list in half. GitHub’s own engineers put retry budgets at the top of their fix list, ahead of three million CPU cores. Yours is a whiteboard.
4. Put one name on it. One person owns the fix. Everyone else holds and does their actual job. Leadership by committee is no leadership at all, and it is never more expensive than during an emergency.

Take it to your team

Read these three out loud in your Monday meeting. Do not send them around. Say them.

• Think about our last emergency. What did each of us add to it, and how much of what we added actually moved the fix forward?
• When something breaks around here, what is the first thing we reach for, and what does reaching for it protect us from feeling?
• If one person had to own our next crisis alone, who would you want it to be, and what should the rest of us be doing instead of helping?

The uncomfortable answer usually shows up in the second one. The pile-on is rarely about the problem. It is about not being able to sit still while something is broken and your name is on it. Which is why knowing how to stay calm under pressure as a leader is not a soft skill. It is what stands between your team and a seven hour outage that should have been a two hour one.

Keep your eyes on the target. — Errol