Why prod is down | Hacker News Reader

Why prod is down

gruntwork.io

54 pointsnapolux12 comments

The underlined placeholders aren't inputs and cannot be filled out.

Also missing are a few options on what to do, including: "Let it burn", "Throw more gasoline at it", and "Quit while you still have some sanity left".

>Worst of all, our database is not backed up, so we lost our data backed up, but the backups don't work, so we lost our data sharded, so we lost our data twice NoSQL, so we're webscale, and we've been losing data for years

Heh. What I keep running into at various companies: "Backed up, but for some reason the restore is going to take three days (if it works), and it will be faster to tap original sources.

Why are we doing backups, then? Company policy.

No one ever got fired for doing backups. It's "industry standard" so if you don't do it, you'll be out of your job faster than the server which went down. Backups being actually restorable in a reasonable amount of time isn't "industry standard", so nobody cares.

Because backing up stuff is super easy, but no one actually tests restoring data until they need to do it live.

We'll do it live! We'll write it and we'll do it live!

It adds to the excitement.

It's scary how relatable it is

My goodness, this is a work of art.

We failed over but there was nothing there so we can't fail back -- eek, we're dead :-)

Somewhat related, has anyone used gruntworks? Thoughts?

Is that real ?

Yes, those are real words written by a real person, stored on a real server and delivered to you by a real Internet.

A better question might be: "Is this really a copy of a letter written by a particular person to another particular person describing a particular incident?" to which the answer is "No."

An even better question might be: "Do the problems outlined in this document really occur?" to which the answer is "Yes, frequently."

Reply on news.ycombinator.com