Fleet dashboard needs:

• Signals into improving the policy. Clusters not using the block cache that should be?
• Errors
• Links to field guide etc, method of providing feedback, DRI for the policy
• Dry-run, running, engaged, disabled, ineligible, unknown
• Number of clusters tuned in selected time frame
• Breakdown of different clusters running the policy by Mongotune version, MongoDB server version, policy configuration (TODO what is this??)
• For each alert, how close to firing it is
• Error count frequencies for poll and tune errors, along with a top-10 list of clusters with highest error frequency, apparently each cluster name needs to be a link to the per-cluster page??
• Last error codes by cluster count
• Something about conflicts of control? To see if Mongotune is fighting an Atlas override?

Per-cluster dashboard needs:

• Brief description of the policy
• Links to: Fleet-wide view, Field guide, Troubleshooting wiki/escalation procedure, Method of providing feedback, DRI for the policy
• Policy state per host (Dry-run indicator, Clear view of when the policy has taken action, Thresholds for decision-making actions)
• Cluster-health signals and how they should be interpreted (Signals that show how the policy is helping, Signals that show how the policy is hurting)
• Conflicts with Atlas custom overrides
• Errors that have occurred: Total error count, Last error code
• Cluster-specific signals that feed policy alerts – threshold to show when an alert will fire
• Descriptions for anything non-obvious
• "Make sure the visualisations provide signal for: dry-run, real engagement, bad configuration, disablement, telemetry gaps, rollback, and recovery scenarios."