# ops-trail

Though we would like to reduce manual operations by automation, there
are still some scenarios that require manual operations.

To reduce the risk, we need to review the manual change steps before
executing it. And then audit the manual changes periodically.

When SRE do manual changes on prod, we need to send a PR containing
the planned manual changes, and execute the manual changes after the
PR is reviewed and merged into `main` branch. Run `./trail.sh` to generate
a new entry for your trial log:

```bash
./trail.sh -m "my great trail"

# you may also specify a trail template, all templates are available in ./trails dir, base.tpl will used by default
./trail.sh -t edit-tc -m "alter cluster configuration of A"
```

You have to fill all the blanks in the generated entry, including:

* Why the manual changes are needed
* How the changes should be made (detailed change steps)
* What is the impact for the manual changes
* What is the recovery operation if the changes fail
* Who will make the changes
* When the changes are planned to be made (time window)