There are several methods I can come up with for this:
- Configuring the replica web server on a different port and / or IP, and then using DNS as a load balancer; restarting one server at a time
- Using more explicit load balancing (which PaaS like Heroku and OpenShift use) with implicit replicas
- Using the built-in mechanism (e.g. in nginx)
I work in an IaaS solution and am creating gitsome listeners to handle this entire installation.
What is the best way to restart the web server so that my latest version of my Python web application can live without much impact on site visitors / users / clients?
source
share