تتبع الأخطاء البرمجية

There's a bug in production.

Can you reproduce it?

It only happens on a page reload.

I can't reproduce it locally.

What are your repro steps for that one?

Do you have the full stack trace from the log?

I'll paste it in the ticket right now.

The errors started after yesterday's deploy.

Let's look at what shipped in that deploy.

I think it's a race condition.

It happens about one in twenty times.

That explains why it's flaky.

I'll add some logging around it.

Found it, it's a null.

We're not checking for it before we use it.

I'll add a guard and a test.

Should we push a hotfix?

Or should we roll it back?

Rolling back is faster and safer.

I'll keep an eye on the logs for an hour.