Allow multiple targets templates files. about ansible-prometheus HOT 11 CLOSED

Just to clarify. After creating those files you'll like to load them with file_sd like this:

  - file_sd_configs:
    - files:
      - /etc/prometheus/file_sd/*.yml

?

from ansible-prometheus.

No, I would like to be able to configure them with a specific job.

See this prometheus config and how we're currently deploying the config files.

from ansible-prometheus.

You are using this role at FOSDEM?! So great! 🎉 Shame I cannot go there this year 😟

But back to the subject. Look how we apply file_sd right now in defaults/main.yml. We can just change line L54 to get wildcard and that way by default it will load all target jobs. However if someone wants to overwrite this behaviour, he can just by specifying different prometheus_scrape_configs. This way role could create custom target files and you could load them with custom jobs. Is this ok?

from ansible-prometheus.

No, using a wildcard there goes against Prometheus job design. It would basically lump all targets into one job. This is why I changed the default scrape_config to only specify one target file. 😃

The goal here is to allow many different jobs with different sets of target lists.

from ansible-prometheus.

I get it. However we need some default way to load those files. Imagine when someone decides to describe targets in prometheus_targets but forgets to write custom prometheus_scrape_config what then? This way we ensure prometheus will load this config albeit not the way it was designed to work.

from ansible-prometheus.

Adjusting scrape configs and targets at the same time is basically a requirement for Prometheus anyway. You can't have targets without a correctly identified scrape config.

from ansible-prometheus.

You are right. However you can define it via label system, look at this: https://github.com/cloudalchemy/demo-site/blob/master/group_vars/all/vars#L9 and then this: http://demo.cloudalchemy.org:9090/targets

One scrape config, multiple jobs.

from ansible-prometheus.

I don't know if such usage is an anti-pattern or isn't, but it's the simplest way we could think of. With such config you just say: here is your target list, they belong to this environment and do this job.

from ansible-prometheus.

The example demo site configuration is not the right pattern to define multiple jobs.

from ansible-prometheus.

Ok, we'll change it after doing #60

from ansible-prometheus.

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.

from ansible-prometheus.