Add config option for log_level

[MichaelThamm]

Closes #188

The addition of the log_level charm config option allows the user to filter classes of logs based on these allowable upstream levels: ["debug", "info", "warn", "error"]. This is useful in the event of a surge of info or debug logs that prevent the user from being able to filter for more important logs like level=error.

TODO

• Add tests
  • Scenario test
    • Scenario dir root tests deleted due to errors running on main.
    • Use consistent naming of context vs. ctx
  • Bump to ops[testing] instead of ops-scenario.
• Set Blocked status and centralised status settings because of model departure but let workload continue.
  • Open an issue in Prom for the same code here.

Testing

1. Deploy and set config

juju deploy zookeeper --trust
juju deploy ./grafana-agent_ubuntu-22.04-amd64.charm gagent --trust
juju relate zookeeper gagent
# Wait for active/idle
juju config gagent log_level=error

Note: If you set an invalid log_level it will default to info and juju debug log will show:

unit-gagent-5: 15:10:07.219 WARNING unit.gagent/5.juju-log Invalid loglevel: panic given, debug/info/warn/error allowed. defaulting to INFO loglevel.

2. Check if the Grafana Agent config file is updated
   juju ssh gagent/0 cat /etc/grafana-agent.yaml | grep log_level
3. Check if the logs are filtered
   juju ssh gagent/0 "journalctl | grep grafana-agent"
4. Scale the principal charm
   juju add-unit zookeeper -n 1
5. Re-run the tests from before in the scaled scenario

• All manual tests passed

[sed-i]

Good job!
A few minor comments.