feat: allow deploying dashboard and alerts on existing instances

[aslafy-z]

This PR includes several changes to allow the use of this Helm chart with existing infrastructure components such as Prometheus Operator and Grafana.

• global.alerts.enabled (new value): enable/disable deployment of alerts
• global.alerts.kind (new value): alerts container resource (configmap or prometheusrule) (Prometheus Operator compatibility)
• global.dashboards.enabled (new value): enable/disable deployment of dashboards
• global.dashboards.labels (new value): allow to set dashboard configmaps a label (Grafana Sidecar compatibility)
• runbookViewer.enabled (new value): enable/disable deployment of the runbook viewer
• Fallback to GitHub-hosted runbook when runbookViewer is disabled
• Only deploy collector configmap when grafana-agent is enabled
• Prefix Grafana dashboards names by Coder
• Prefix Grafana dashboards configmap names and Grafana uid (posponed Make Grafana dashboards' UIDs consistent #63) by coder-

It does not fix #31 but helps people that only want to deploy dashboards and alerts for example.

I'm running it right now with these values and it works almost perfectly. Without the grafana agent, logs are not correctly matched (logger field not detected) - I've seen json support was tracked with #8, it may resolve this issue. I had to add service monitors by myself, I opened issue #67 for this matter.

demo v0.6.0 is available in the https://zadkiel.fr/coder-observability helm repository

[dannykopping]

@aslafy-z thanks very much for your contribution. I'll be on PTO until Mon 13th, and my colleague who is co-maintaining this with me is away as well. Would it be ok with you if I got to this when I return?

[aslafy-z]

@dannykopping sure, enjoy your pto!

[dannykopping]

Overall LGTM! Thanks for a great contribution @aslafy-z
Left a few notes

[aslafy-z]

@dannykopping I rebased and re-built the compiled manifests. Do you see any blocking things left? Thanks

[dannykopping]

@aslafy-z the main sticking point for me is #62 (comment)
Have you tested this yet?

[aslafy-z]

Sure! I missed that comment. I will test it in my dev environment in a few hours and will get back to you.

[aslafy-z]

global:
  # global.postgres -- postgres connection information
  postgres:
    hostname: postgresql-cluster
    port: 5432
    username: coder
    database: coder
    sslmode: require
    mountSecret: coder-postgres-config
    exporter:
      enabled: true
      image: quay.io/prometheuscommunity/postgres-exporter:v0.18.1

  alerts:
    # global.alerts.kind -- the container resource kind in which alerts should be created;
    # valid values are "prometheusrule" or "configmap"
    kind: prometheusrule

  dashboards:
    enabled: true
    # global.dashboards.labels -- labels to apply to configmaps created for dashboards
    labels:
      grafana_dashboard: "1"

runbookViewer:
  enabled: false

sqlExporter:
  enabled: false

grafana-agent:
  enabled: false

grafana:
  enabled: false

prometheus:
  enabled: false

pyroscope:
  enabled: false

loki:
  enabled: false

tempo:
  enabled: false

I'm running it right now with these values and it works almost perfectly. Without the grafana agent, logs are not correctly matched (logger field not detected) - I've seen json support was tracked with #8, it may resolve this issue. I had to add service monitors by myself, I opened issue #67 for this matter.

demo v0.6.0 is available in the https://zadkiel.fr/coder-observability helm repository

[dannykopping]

Almost there, thank you for your patience @aslafy-z - it's greatly appreciated!

[dannykopping]

Thanks for all the hard work on this @aslafy-z! Much appreciated.

[aslafy-z]

@dannykopping would you mind labelling this PR with hacktoberfest-accepted so it could be counted as eligible for my Hacktoberfest PR count? Thank you

[dannykopping]

@aslafy-z done!