Last week we launched the new website for 18-55 Productions!
18-55 is a video production company, based in Bordeaux. They work with various brands such as DC Shoes, Orange, or even Electronic Arts amongst many others. Their goal was to get a website that could put a focus on the content (their productions) and at the same time show that they also act as a platform for different type of professions revolving around their line of work. That’s why they’re surrounded by various artists, photographers, writers and videographers which allow them to quickly put up a team of talented people for the projects they’re working on.
Their second desire was to get a back-office which would allow them to add/remove content quickly and easely. WordPress was the obvious choice, and together we came up with a minimalistic design based on a grid of squares which allowed them to reorder the whole site as they please. They were immediately thrilled by this approach and after a few exchanges between their CEO, the Production Manager and myself we agreed on this design. All in all I’m really happy with what we came up with and so far the feedback after launch is rather positive.
Thanks a lot to 18-55 for their time and dedication to this project!
A couple of months ago I bought myself a Pebble. It’s a nifty little device which acts as an extension to my phone, directly strapped to my wrist. Thanks to it I can quickly send pre-recorded messages with the press of a few buttons, but most importantly get notifications on new messages, them being emails, SMS or even Telegram messages. Which can get a bit invasive especially when my friend Edouard spams me with Wu-Tang Clan lyrics. All in all, it’s really an amazing gadget and I’m really happy with it.
Another cool feature is its ability to track sleeping patterns. Thanks to its built-in gyroscope, it can detect wrist movements and deduce if I’m sleeping or not. Granted, it’s not 100% bullet-proof, but it does a good job at giving a broad idea of how good (or bad) my sleeping habits are. And boy they are terrible.
According to my Pebble, I sleep 7 hours 3 minutes per night on a 30 days average, with an average deep sleep duration of 1h 48m. It’s not that bad, but when you take into account that during a same week I can sleep from 3h 55m one day to 9h 18m another, there’s clearly something really wrong with my sleep schedule.
My usual “worst” night is the one from sunday to monday, with an average of 5 hours of sleep. My “best” night is usually the one from saturday to sunday, with up to 11 hours (!) of sleep.
The average human needs between 7 and 8 hours of sleep for optimal functions, which means I have a sleep debt of about 30 minutes to an hour per day. That’s a cumulated debt of about 4-5 hours per week. On the long run, such deprivation is the cause of stress, weight gain, fatigue, and even diabetes.
“But Fabien, all you need to do is go to sleep one hour earlier, that’s all!”
Yeah, that’s not taking into account the absolutely amazing capacity of the human brain to slack off.
Getting more sleep has been a goal of mine for 2017. I’ve already taken some steps towards this, removing my tablet from the nightstand and replacing it with a good book. But I still need to slap myself on the wrist and stop spending my evenings reading about code or cruising around Tamriel or Dust2.
I hope to write an addendum to this post soon, describing how my sleep habits have changed, if I ever manage to change them. In the mean time, I’m off to take a nap.
Cover image: Dominic QN.
Last night, a “tired sysadmin” ran a directory wipe on the wrong server of code hosting platform Gitlab, deleting over 300 GB of live production data. Following that incident, the website was taken offline while they were trying to restore from a backup. Plot twist: “none [of their] backup/replication techniques [were] working reliably or set up in the first place.” It took them a full day to fix the damages, leaving thousands of devs stranded, and losing close to 6 hours of data.
On Twitter, Gitlab posted the most amazing and devastating tweet a sysadmin could ever write in their career:
We accidentally deleted production data and might have to restore from backup.
To be honest, there couldn’t be a more disastrous catastrophy for a company that’s relying on data as much as they do. Me and the team I’m currently working with have been using Gitlab for a few months now, and already had to suffer from their constant downtime. At one point, Gitlab.com could be offline three to five times a day. We were glad to hear that they were looking to leave the cloud, only to learn that finally, they’d be looking to expand in the cloud after all.
As Hacker News user connorshea pointed out:
it’s never the fault of a person, always of a system
This could be applied to what happened at Gitlab. It feels amazing that for such critical tasks (you don’t delete a whole database directory every day), they don’t rely to the most simple and effective management system: checklists. It is used around the globe whenever a system is at risk. Airline pilots, surgeons and nurses, army crew, they all have checklists. It is the safenet against repetitive and menial tasks. That’s why you’ll see airline pilots talking through everything they’re doing during an emergency procedure, with their co-pilot enforcing what they’re about to do. It usually goes like this:
– “I’m about to turn off A/C.
– Green for turning off A/C.
– A/C is turned off.
– Confirmed. A/C is turned off”.
This is greatly exagerated as I haven’t flown a plane in.. well ever. But you get my point. Each and every step is monitored and confirmed by a fellow crewman. If you want to know more you can read this Wikipedia article about crew resource management.
Gitlab has taken the stance of total transparency, going as far as live-broadcasting their backup process. Many praise them for taking such a stance, but they hardly couldn’t take another one, as there was a slight possibility of irrecuperable failure.
At the end of the day, the real question is not really “who‘s to blame?” but rather “what‘s to blame?”. Poor practices, lack of preparation and a frustrated and tired engineer make for a bad combo. I’m sure Gitlab will grow from its mistakes. I’m not so sure they’ve handled the situation perfectly and are not losing clients by the minute, but hey, we can’t blame them from trying not to go under.