Skip to content
Corentin GS

Begin Again Without Pretending Nothing Happened

№45 · · ·1064 words ·5 min read
Begin Again Without Pretending Nothing Happened
In this piece

At 6 p.m., a client called because their production site was down.

The team was panicking. They were talking about larger servers before they had identified the failure or contained its impact. Every service was hitting the database. A burst of traffic had triggered too many transaction inserts, lock congestion had spread through PostgreSQL, and the website could no longer respond. There was no circuit breaker, monitoring, or observability to show where the failure began or how far it had reached.

I found the Lambda that was creating the inserts and shut it down. Data began to pile up in its queue. That was the right trade: the queue absorbed the pressure while the database recovered. The site came back. I reduced the Lambda concurrency to a few workers so the queue, rather than the database, would take the next surge.

The incident needed more than that immediate fix. We still had to add the missing safeguards and understand the failures in the infrastructure and the code. I spent the night reading PostgreSQL logs and Datadog dashboards, tracing each condition that had turned traffic into a full outage. The technical post-mortem deserves its own devlog.

When the system was stable, I felt guilty. I had not created the failure, and the affected Lambda was outside my original scope. Still, I was the consultant expected to notice the risks nobody had asked about. I should have seen the missing controls and raised the concern earlier.

Closing the terminal windows felt too easy, almost dishonest. The damage had happened. People had lost time. A clean screen could not make that untrue.

Carrying the incident forward only extended the emergency into the next task. It repaired nothing.

The question was:

How do I mark an ending without pretending nothing happened?

A resolved incident workspace on one side of a threshold, with a clear desk and notebook ready for the next task on the other

What Purification Restores

While reading about Shinto, I kept returning to harae, commonly translated as purification. Kokugakuin University’s Encyclopedia of Shinto entry on harae defines it as purifying mind and body of accumulated sins and defilements through ablutions, rites, or recitations.

Purification can sound like erasure, as though cleaning removed the truth of what happened. Cleaning a wound acknowledges the injury and treats what remains. Clearing a damaged workspace serves the same purpose: respond to the residue.

Mistakes leave information and residue: a changed environment, a defensive posture, unfinished conversations, and habits formed under pressure. A debrief can identify the lesson while the residue remains.

Part 1 gives the operational work its home: reconstruct the event and carry the lesson forward. This article begins after that work, when understanding has arrived but readiness has not.

Residue, Guilt, and Repair

The language of impurity becomes dangerous when someone turns it into a judgment about people. I use residue without calling anyone dirty, damaged, or morally lesser. Shinto concepts have histories and ritual meanings far beyond my psychological analogy.

Here, residue names what I can observe after a difficult event: the open terminal, the clenched jaw, the urge to check the same dashboard again, the apology still owed, or the procedure still waiting to change.

Some residue calls for repair. If I harmed someone, I need to acknowledge it. If the process invited the failure, the process needs work. If data remains at risk, cleanup is part of the incident. No gesture can replace those obligations.

Other residue belongs to the transition itself. The emergency required speed, interruption, and a narrow field of attention. The next piece of work may require patience and curiosity. Moving between those states requires a deliberate threshold.

The discipline of returning helps me bring attention back when regret takes it away. A physical threshold gives that return a place and a time.

Why a Threshold Helps

We already use thresholds without giving them much thought. A chess player resets the pieces after a game. An engineer closes the incident channel and returns the dashboards to their usual state.

These actions signal that one mode of attention has ended.

Without a threshold, tasks bleed into each other. I answer an ordinary message with the urgency of an outage. I review a teammate’s code while still defending my own mistake. I open a new document, but part of me remains in the old terminal.

A threshold creates a point at which I can ask whether I am ready to continue and what must happen first.

Sometimes the answer is “not yet.” The system may need another check. A conversation may still be open. I may need sleep before making a decision. Marking the threshold reveals unfinished work that the next task would otherwise bury.

A four-step reset after an incident: Close, Clean, Repair, and Cross

My Reset After an Incident

I use a short reset after a failure or a period of intense work. This is my personal practice, inspired by what I have read about purification and thresholds. It is not a Shinto ritual, and I would not present it as one.

Close. I close the incident tools and put away the notes that belong to the event. Before doing so, I confirm that the facts, owners, and next actions have a durable home.

Clean. I wash the cup, clear the desk, and open a window when I can. The actions are ordinary on purpose. I want the workspace to match the work I am about to do.

Repair. I write down any obligation that remains: an apology, a follow-up, a rollback check, or a change to the process. If I cannot name where it will be handled, the incident is not ready to close.

Cross. I step away for a few minutes. When I return, I name the next task before opening anything. This keeps the loudest leftover signal from choosing by default.

The practice takes less than ten minutes. Its value comes from repetition. The same actions mark the same change: emergency attention has ended; deliberate attention can begin.

Beginning Again Honestly

A clean desk says nothing about my innocence, and repairing trust takes more than an open window. The reset earns its place only after responsibility has somewhere concrete to go.

That is why I keep the written follow-up before I clear the physical traces. I keep the facts, commitments, and lesson somewhere I can find them.

Then I can let the emergency posture end.

Beginning again moves toward a new state that includes the repair, the lesson, and the knowledge that another failure is possible.

Keep the past in the record, and leave the next action clear.

Next in The Next Move

Learn Backward. Act Forward.

Part 4 continues the series.