11 ms·

There's a difference between "hosting all of Eats Canada" and "hosting some component or service until we can fix prod." I think it's more likely similar to the

There's a difference between "hosting all of Eats Canada" and "hosting some component or service until we can fix prod." I think it's more likely similar to the latter

dvrp 3y ago
That was my thoughts as well. Maybe someone from Uber Eats Canada can clarify.
whalesalad 3y ago
while true; do ssh intern@eats.uber.ca 'find /var/log -type f -delete'; sleep 300; done
bpicolo 3y ago
This tracks. At a former employer, one of their acquired startups (that I joined on to help guide technical direction for) ran their entire order flow by calling a url every minute via a cron task from the primary database server. At one point, the newer devops organization migrated to a bigger, beefier MySQL (this is when fast NVMe was first available in specialized EC2 instances). Nobody remembered to migrate the cron script at that moment, so for about an hour I ran all of customer order flow with a bash script curling a public facing (yikes) url to initialize those jobs.
adrr 3y ago
Probably was running a script that ssh'ed into production boxes and restart some service at some specified interval or cleared some log files to stop the drive from filling up.
water-data-dude 3y ago
“until we can fix prod”, but then other stuff comes up, and just look at the backlog! And, I mean, it works pretty well for now…and before you know it, it’s been there forever