@nixCraft

7. Always transactions when you can. Yikes.

@nixCraft I didn't get root privileges on a Unix (like) box until I was about 35, so it took me a little while to catch up.

There are a couple of those I've done (again) within just the few months.

@nixCraft

7, in production, about 25 years ago.

All the others were just Tuesdays for me.

@nixCraft I did a

get /lib/libc.so

once, using ftp from an x86 system to an arm system. As root. Fond memories.

@nixCraft All except 3 and 5.
I actually make backups from important data.

4 was the proxy server of the main office. Only took down all inter office communication for like 2 minutes *grin*
1 not really but I did dd to the wrong disk, does that count?

@nixCraft

Mhmh.. strangely I never did the rm -rf / thing.
But I'm good at damaging systems by mistake.
I say I'm doing manual chaos engineering, so we can test our recovery plans.

@nixCraft

I can tick all those except 3 and 7. Dam, I'm a failure—must try harder 😔

@nixCraft

1. sudo rm -rf --no-preserve-root
2. kubectl delete namespace (not me, a junior)
3. Refused to reboot firewall servers for updates for years "because it would reset uptime" (my senior when I was much younger)
4. UPDATE across all rows without a transaction to roll back, and the incorrect SET, basically nulling out all data
5. Thought I was in a subdirectory, and ran "rm * -rf" (luckily, the .git directory persisted, so a "git reset --hard" fixed things)

@nixCraft

It was a really nice morning so far, thanks for the war flashbacks 😅

✅ 2. Reboot production server by mistake
✅ 5. No backup was made/verified ever...
✅ 6. Delete source code before actually committing to source control
✅ 7. Run a SQL update/delete with wrong where clause

@nixCraft Fuck, I am 40+.

And I am still missing the first three...

@nixCraft done or witnessed all of them.

Best one was trying to explain how FTP worked to a minion at a multinational shipping and postage company. They screwed up their firewall on the call and locked everyone out of their production system in the middle of the working day. Half an hour later the VP called me back to apologise and say all was well again. We never had a problem again with them after that...

You live and learn...

@nixCraft * without a Where clause

Does this count ?

@nixCraft
Never did 3, and can't remember if I ever did 5 or 7.

And 1 was intentional.

4 was fun .. telling a non techie in a very remote office what to type - symbol for symbol - into a local terminal to repair the problem afterwards.

@nixCraft ah, that 7. one got me hard once

didn't even notice that I deleted like 1/10th of the production DB

went on vacation

when I came back the backup was overwritten already

but it was only about some stupid real estate trading thing… something-something a billion euro per year in trades.

nothing to worry about!

@nixCraft @enobacon *old man shakes fist at clouds* kids these days have it so easy! There aren’t enough hardware mistakes in this list!

And if you only rm -rf / on one machine, you’re not doing *real* #DevOps