Quote:
Originally Posted by POBEP
My best guess would be a deployment or backup or both.
Someone (poor) design that does not allow for life deployments.
For example, design may call for syncing source code across pool of existing web servers.
During this time, the "code" is inconsistent, hence poor choice is
- hold traffic
- sync source
- restart webapp
- resume traffic
- hope it works
I work at the place like this once.
When I started - 20 min "deploy outage" was norm.
6 weeks later I rolled out very simple improvement that reduced "scheduled outage" from 20 minutes to 1.5 seconds.
Solution was simple:
- deploy new code side-by-side (or on new infra)
- flip the dns and or activation symlink on the file system
Guess what? Customers noticed - employees and management did not
|
Yeah it should be something like that. It should be pretty simple to fix the website. Since it is automated anyway, they should try to push the time to somewhere later at night or hire me to fix the script lol