How sentiment powers triage
Sentiment turns the dashboard into a focus list. Every KPI is sorted into one of three states you can filter on:- Performing - comfortably inside the good band.
- Needs Attention - has breached the band you set. This is where to act.
- Stable - within range, or an informational card with no good/bad direction.
| Card | Direction | Good | Bad | Unit |
|---|---|---|---|---|
| Connection Pool Saturation % | Lower is better | 70 | 90 | % |
| Connection Pool at >90% Saturation | Lower is better | 70 | 90 | Count |
| Database Disk Usage % | Lower is better | 70 | 90 | % |
| Last Successful Backup (hours ago) | Lower is better | 24 | 72 | Count |
| MongoDB Pool Saturation vs Traffic Burst | Lower is better | 70 | 90 | % |
| Query Error Rate % | Lower is better | 0.1 | 1 | % |
| Query Error Rate Spike (>1% in 5m) | Lower is better | 0.1 | 1 | Count |
| Query Latency p95 (ms) | Lower is better | 50 | 200 | Count |
| Query Latency p99 (ms) | Lower is better | 100 | 500 | Count |
| Replica Lag (seconds) | Lower is better | 1 | 10 | Count |
| Replica Set Member Lag >10s or in RECOVERING State | Lower is better | 1 | 10 | Count |
| Slow Ops (15m, >100ms) | Lower is better | 1 | 5 | Count |
| WiredTiger Cache Hit Rate % | Higher is better | 95 | 80 | % |