Mitigating the risks of automatic deployment

Deployment

I currently work for a company that deploys through github. However, we need to log in to all 3 servers to update them manually using the shell script. When he spoke with the CTO, he made clear that automatic deployment was like voodoo. Which is understandable. We have developers from 4 different countries working remotely. If someone accidentally gets into the wrong branch, we can experience downtime, and with our service we will not be able to sleep for more than 10 minutes. And with all our developers at different times, our technical director will not know until the next morning, and it will be difficult for us to meet developers who have had a problem due to the huge time differences.

My problem: why do I want automatic deployment

While working on my personal project, I decided it might be in my interest to use automatic deployment, but still my project is critical, and I would like to reduce downtime and human errors as much as possible. The problem with manual deployment is that I simply cannot manually deploy up to 20 servers via SSH in a reasonable amount of time. The problem persists when I consider autoscaling. I will need to unwind a new server from the image and deploy it.

My stack

Node.js Express. . npm package.json uglify , forever-monitor. grunt.js, .

:

  • git, -
  • git
  • npm
  • Docker

, Docker, , , , Docker, . .

:

, , . - -, , .

, . . .

,

, , , . - .

0

All Articles