mirror of
https://github.com/Memo-2023/mana-monorepo.git
synced 2026-05-14 20:41:09 +02:00
- Replace Prometheus with VictoriaMetrics (2-year retention) - Add DuckDB analytics module for business KPIs (unlimited retention) - Add master overview dashboard combining all metrics - Add business metrics dashboard for user growth tracking - Add backup script for VictoriaMetrics snapshots and DuckDB - Add ADR documentation for monitoring stack decision Analytics API endpoints: - GET /api/v1/analytics/health - Service health - GET /api/v1/analytics/latest - Latest metrics snapshot - GET /api/v1/analytics/growth - User growth over time - GET /api/v1/analytics/monthly - Monthly aggregates - POST /api/v1/analytics/snapshot - Manual snapshot trigger
593 lines
20 KiB
Markdown
593 lines
20 KiB
Markdown
# ADR-001: Monitoring Stack Upgrade - VictoriaMetrics + DuckDB
|
|
|
|
**Status:** Accepted
|
|
**Date:** 2025-01-28
|
|
**Author:** Till Schneider
|
|
**Reviewers:** -
|
|
|
|
## Executive Summary
|
|
|
|
Upgrade des ManaCore Monitoring Stacks von Prometheus (30 Tage Retention) auf VictoriaMetrics (2 Jahre) + DuckDB (unbegrenzt) für langfristige Metriken-Speicherung und Business-Analytics.
|
|
|
|
---
|
|
|
|
## 1. Kontext & Problemstellung
|
|
|
|
### 1.1 Aktuelle Situation
|
|
|
|
ManaCore nutzt einen Standard-Prometheus + Grafana Stack für Monitoring:
|
|
|
|
```
|
|
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
|
|
│ NestJS Backends│────>│ Prometheus │────>│ Grafana │
|
|
│ (6 Services) │ │ (30 Tage) │ │ (5 Dashboards) │
|
|
└─────────────────┘ └─────────────────┘ └─────────────────┘
|
|
│
|
|
├── mana-core-auth (Port 3001)
|
|
├── chat-backend (Port 3002)
|
|
├── todo-backend (Port 3018)
|
|
├── calendar-backend (Port 3016)
|
|
├── clock-backend (Port 3017)
|
|
└── contacts-backend (Port 3015)
|
|
```
|
|
|
|
**Komponenten:**
|
|
- Prometheus v2.51.0 mit 30 Tagen Retention
|
|
- Grafana 10.4.1 mit 5 Dashboards
|
|
- Node Exporter, cAdvisor, PostgreSQL Exporter, Redis Exporter
|
|
- Alerting Rules (20+ Regeln)
|
|
|
|
### 1.2 Das Problem
|
|
|
|
**Nach 30 Tagen sind alle historischen Metriken unwiederbringlich verloren.**
|
|
|
|
| Betroffene Daten | Konsequenz |
|
|
|------------------|------------|
|
|
| User-Wachstum (`auth_users_total`) | Keine Trend-Analyse möglich |
|
|
| Historische Error Rates | Keine Langzeit-Vergleiche |
|
|
| Performance-Trends | Keine Kapazitätsplanung |
|
|
| Infrastruktur-Metriken | Keine saisonalen Muster erkennbar |
|
|
|
|
**Besonders kritisch:** Business-KPIs wie `auth_users_total`, `auth_users_created_this_month` sind Point-in-Time Snapshots. Ohne historische Daten ist es unmöglich zu rekonstruieren, wie viele User vor 2 Monaten existierten.
|
|
|
|
### 1.3 Anforderungen
|
|
|
|
| Anforderung | Priorität |
|
|
|-------------|-----------|
|
|
| Operative Metriken für mindestens 1-2 Jahre speichern | Hoch |
|
|
| Business-KPIs unbegrenzt speichern | Hoch |
|
|
| Keine Änderung an bestehenden Dashboards | Mittel |
|
|
| Minimaler zusätzlicher Ressourcenverbrauch | Mittel |
|
|
| Einfache Wartung und Backup | Mittel |
|
|
|
|
---
|
|
|
|
## 2. Evaluierte Optionen
|
|
|
|
### 2.1 Option A: Prometheus Retention erhöhen
|
|
|
|
**Ansatz:** `--storage.tsdb.retention.time=365d`
|
|
|
|
**Vorteile:**
|
|
- Keine Migration nötig
|
|
- Keine neuen Komponenten
|
|
|
|
**Nachteile:**
|
|
- Prometheus TSDB ist nicht für Langzeit optimiert
|
|
- RAM-Verbrauch steigt linear mit Retention
|
|
- Queries über alte Daten werden langsam
|
|
- Compaction-Overhead bei großen Datenmengen
|
|
|
|
**Bewertung:** Kurzfristige Lösung, skaliert nicht.
|
|
|
|
### 2.2 Option B: Thanos / Cortex
|
|
|
|
**Ansatz:** Prometheus + Langzeit-Storage-Layer (S3/MinIO)
|
|
|
|
**Vorteile:**
|
|
- Industriestandard für große Deployments
|
|
- Unbegrenzte Retention möglich
|
|
|
|
**Nachteile:**
|
|
- Hohe Komplexität (5+ zusätzliche Komponenten)
|
|
- Overkill für ManaCore's Größe (~50k Time Series)
|
|
- Signifikanter Ops-Overhead
|
|
|
|
**Bewertung:** Overengineered für unseren Use Case.
|
|
|
|
### 2.3 Option C: VictoriaMetrics (gewählt)
|
|
|
|
**Ansatz:** Drop-in Replacement für Prometheus
|
|
|
|
**Vorteile:**
|
|
- 100% Prometheus-kompatibel (PromQL, Config-Format, Exporters)
|
|
- 3-10x bessere Kompression
|
|
- 5-10x weniger RAM-Verbrauch
|
|
- Schnellere Queries über historische Daten
|
|
- Single Binary, einfaches Deployment
|
|
- Migration in 10 Minuten
|
|
|
|
**Nachteile:**
|
|
- Weniger bekannt als Prometheus (aber wachsende Community)
|
|
- CNCF Sandbox (nicht Graduated wie Prometheus)
|
|
|
|
**Bewertung:** Beste Balance aus Einfachheit und Leistung.
|
|
|
|
### 2.4 Option D: PostgreSQL für Business-Metriken
|
|
|
|
**Ansatz:** Tägliche Snapshots in PostgreSQL speichern
|
|
|
|
**Vorteile:**
|
|
- Bestehende Infrastruktur nutzen
|
|
- SQL für Queries
|
|
- Unbegrenzte Retention
|
|
|
|
**Nachteile:**
|
|
- Nicht optimiert für Analytics-Queries
|
|
- Connection-Pool Overhead
|
|
- Row-based Storage ineffizient für Aggregationen
|
|
|
|
**Bewertung:** Funktional, aber nicht optimal für Analytics.
|
|
|
|
### 2.5 Option E: DuckDB für Business-Metriken (gewählt)
|
|
|
|
**Ansatz:** Embedded OLAP-Datenbank für tägliche Business-KPI Snapshots
|
|
|
|
**Vorteile:**
|
|
- Kein Server nötig (embedded, single file)
|
|
- Column-oriented = perfekt für Analytics
|
|
- 10-100x schneller als PostgreSQL für Aggregationen
|
|
- Exzellente Kompression
|
|
- Native Parquet Import/Export
|
|
- SQL-kompatibel
|
|
|
|
**Nachteile:**
|
|
- Nicht für concurrent writes (irrelevant bei 1x täglich)
|
|
- Keine native Grafana-Integration (API-Endpoint nötig)
|
|
|
|
**Bewertung:** Perfekt für den Use Case (append-only, read-heavy, analytics).
|
|
|
|
---
|
|
|
|
## 3. Entscheidung
|
|
|
|
### 3.1 Gewählte Architektur
|
|
|
|
```
|
|
┌─────────────────────────────────────────────────────────────────────────┐
|
|
│ ManaCore Monitoring Stack v2 │
|
|
├─────────────────────────────────────────────────────────────────────────┤
|
|
│ │
|
|
│ OPERATIVE METRIKEN (High-Frequency Time Series) │
|
|
│ ════════════════════════════════════════════════ │
|
|
│ │
|
|
│ ┌──────────────┐ ┌──────────────────┐ ┌─────────────────┐ │
|
|
│ │ Backends │────>│ VictoriaMetrics │────>│ Grafana │ │
|
|
│ │ /metrics │ │ │ │ │ │
|
|
│ │ │ │ Retention: 2y │ │ Existing │ │
|
|
│ │ + Exporters │ │ Scrape: 15-30s │ │ Dashboards │ │
|
|
│ └──────────────┘ └──────────────────┘ └─────────────────┘ │
|
|
│ ▲ │
|
|
│ │ │
|
|
│ BUSINESS METRIKEN (Daily Snapshots, Analytics) │ │
|
|
│ ══════════════════════════════════════════════ │ │
|
|
│ │ │
|
|
│ ┌──────────────┐ ┌──────────────────┐ │ │
|
|
│ │ Daily Cron │────>│ DuckDB │────────────┘ │
|
|
│ │ 00:00 UTC │ │ │ (via JSON API) │
|
|
│ │ │ │ Retention: ∞ │ │
|
|
│ │ Snapshots: │ │ File: metrics.db│ │
|
|
│ │ - Users │ │ Size: ~10MB/year│ │
|
|
│ │ - Growth │ │ │ │
|
|
│ │ - Features │ │ Backup: cp file │ │
|
|
│ └──────────────┘ └──────────────────┘ │
|
|
│ │ │
|
|
│ ▼ │
|
|
│ ┌──────────────┐ │
|
|
│ │Parquet Export│ │
|
|
│ │ (Archiv) │ │
|
|
│ └──────────────┘ │
|
|
│ │
|
|
└─────────────────────────────────────────────────────────────────────────┘
|
|
```
|
|
|
|
### 3.2 Daten-Aufteilung
|
|
|
|
| Datentyp | Storage | Retention | Grund |
|
|
|----------|---------|-----------|-------|
|
|
| CPU, Memory, Disk | VictoriaMetrics | 2 Jahre | High-frequency, Time-Series |
|
|
| HTTP Requests, Latency | VictoriaMetrics | 2 Jahre | High-frequency, PromQL |
|
|
| Error Rates, Status Codes | VictoriaMetrics | 2 Jahre | Alerting, Debugging |
|
|
| Container Metrics | VictoriaMetrics | 2 Jahre | Kapazitätsplanung |
|
|
| **User Counts** | DuckDB | Unbegrenzt | Business KPI, Trend-Analyse |
|
|
| **User Growth** | DuckDB | Unbegrenzt | Business KPI |
|
|
| **Feature Usage** | DuckDB | Unbegrenzt | Product Analytics |
|
|
| **Revenue/Subscriptions** | DuckDB | Unbegrenzt | Business KPI |
|
|
|
|
### 3.3 Warum diese Kombination?
|
|
|
|
**VictoriaMetrics für operative Metriken:**
|
|
- Prometheus-kompatibel = keine Dashboard-Änderungen
|
|
- 2 Jahre Retention bei ~15GB Storage
|
|
- Schnelle Queries auch über historische Daten
|
|
- Bewährte Time-Series Datenbank
|
|
|
|
**DuckDB für Business-Metriken:**
|
|
- Perfekt für "1x täglich schreiben, oft lesen"
|
|
- SQL für komplexe Analytics-Queries
|
|
- Single-File = triviales Backup
|
|
- Kein zusätzlicher Server/Container
|
|
- Unbegrenzte Retention bei minimalem Footprint
|
|
|
|
---
|
|
|
|
## 4. Technische Details
|
|
|
|
### 4.1 VictoriaMetrics Konfiguration
|
|
|
|
```yaml
|
|
# docker-compose.macmini.yml
|
|
services:
|
|
victoriametrics:
|
|
image: victoriametrics/victoria-metrics:v1.99.0
|
|
container_name: victoriametrics
|
|
restart: unless-stopped
|
|
command:
|
|
- '-storageDataPath=/storage'
|
|
- '-retentionPeriod=2y'
|
|
- '-httpListenAddr=:8428'
|
|
- '-promscrape.config=/etc/prometheus/prometheus.yml'
|
|
- '-promscrape.config.strictParse=false'
|
|
volumes:
|
|
- vm-storage:/storage
|
|
- ./prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
|
|
- ./prometheus/alerts.yml:/etc/prometheus/alerts.yml:ro
|
|
ports:
|
|
- "8428:8428"
|
|
networks:
|
|
- manacore-network
|
|
```
|
|
|
|
**Ressourcen-Vergleich (geschätzt):**
|
|
|
|
| Metrik | Prometheus (30d) | VictoriaMetrics (2y) |
|
|
|--------|------------------|----------------------|
|
|
| RAM | ~2 GB | ~500 MB |
|
|
| Disk | ~5 GB | ~15 GB |
|
|
| CPU | Höher (Compaction) | Niedriger |
|
|
|
|
### 4.2 DuckDB Schema
|
|
|
|
```sql
|
|
-- Haupt-Tabelle für tägliche Snapshots
|
|
CREATE TABLE daily_metrics (
|
|
date DATE PRIMARY KEY,
|
|
|
|
-- User Metrics
|
|
total_users INTEGER NOT NULL,
|
|
verified_users INTEGER NOT NULL,
|
|
new_users_today INTEGER NOT NULL,
|
|
new_users_week INTEGER NOT NULL,
|
|
new_users_month INTEGER NOT NULL,
|
|
|
|
-- Engagement (Platzhalter für Zukunft)
|
|
daily_active_users INTEGER,
|
|
weekly_active_users INTEGER,
|
|
monthly_active_users INTEGER,
|
|
|
|
-- Per-App Metrics (Platzhalter)
|
|
chat_messages_sent INTEGER,
|
|
pictures_generated INTEGER,
|
|
|
|
-- Infrastructure Snapshots
|
|
total_db_size_bytes BIGINT,
|
|
total_storage_size_bytes BIGINT,
|
|
|
|
-- Metadata
|
|
recorded_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
|
|
);
|
|
|
|
-- Index für schnelle Range-Queries
|
|
CREATE INDEX idx_daily_metrics_date ON daily_metrics(date);
|
|
|
|
-- View für monatliche Aggregation
|
|
CREATE VIEW monthly_metrics AS
|
|
SELECT
|
|
DATE_TRUNC('month', date) AS month,
|
|
MAX(total_users) AS total_users_eom,
|
|
SUM(new_users_today) AS new_users,
|
|
AVG(daily_active_users) AS avg_dau
|
|
FROM daily_metrics
|
|
GROUP BY DATE_TRUNC('month', date);
|
|
```
|
|
|
|
### 4.3 DuckDB Service Implementation
|
|
|
|
```typescript
|
|
// services/mana-core-auth/src/analytics/analytics.service.ts
|
|
@Injectable()
|
|
export class AnalyticsService {
|
|
private db: Database;
|
|
|
|
constructor(
|
|
private readonly usersService: UsersService,
|
|
private readonly configService: ConfigService,
|
|
) {
|
|
const dbPath = this.configService.get('DUCKDB_PATH', '/data/metrics.duckdb');
|
|
this.db = new Database(dbPath);
|
|
this.initializeSchema();
|
|
}
|
|
|
|
@Cron('0 0 * * *') // Täglich um Mitternacht UTC
|
|
async recordDailySnapshot(): Promise<void> {
|
|
const today = new Date().toISOString().split('T')[0];
|
|
|
|
const metrics = {
|
|
date: today,
|
|
total_users: await this.usersService.countTotal(),
|
|
verified_users: await this.usersService.countVerified(),
|
|
new_users_today: await this.usersService.countCreatedToday(),
|
|
new_users_week: await this.usersService.countCreatedThisWeek(),
|
|
new_users_month: await this.usersService.countCreatedThisMonth(),
|
|
total_db_size_bytes: await this.getDbSize(),
|
|
};
|
|
|
|
this.db.run(`
|
|
INSERT OR REPLACE INTO daily_metrics
|
|
(date, total_users, verified_users, new_users_today,
|
|
new_users_week, new_users_month, total_db_size_bytes)
|
|
VALUES (?, ?, ?, ?, ?, ?, ?)
|
|
`, [
|
|
metrics.date,
|
|
metrics.total_users,
|
|
metrics.verified_users,
|
|
metrics.new_users_today,
|
|
metrics.new_users_week,
|
|
metrics.new_users_month,
|
|
metrics.total_db_size_bytes,
|
|
]);
|
|
}
|
|
|
|
async getUserGrowth(months: number = 12): Promise<GrowthData[]> {
|
|
return this.db.all(`
|
|
SELECT
|
|
date,
|
|
total_users,
|
|
total_users - LAG(total_users) OVER (ORDER BY date) as growth
|
|
FROM daily_metrics
|
|
WHERE date > CURRENT_DATE - INTERVAL '${months} months'
|
|
ORDER BY date
|
|
`);
|
|
}
|
|
}
|
|
```
|
|
|
|
### 4.4 Grafana Integration
|
|
|
|
**VictoriaMetrics:**
|
|
```yaml
|
|
# docker/grafana/provisioning/datasources/prometheus.yml
|
|
apiVersion: 1
|
|
datasources:
|
|
- name: Prometheus
|
|
type: prometheus
|
|
url: http://victoriametrics:8428 # Nur URL ändert sich
|
|
isDefault: true
|
|
editable: false
|
|
```
|
|
|
|
**DuckDB (via JSON API):**
|
|
```yaml
|
|
# docker/grafana/provisioning/datasources/duckdb.yml
|
|
apiVersion: 1
|
|
datasources:
|
|
- name: Business Metrics
|
|
type: simpod-json-datasource
|
|
url: http://mana-core-auth:3001/api/analytics
|
|
isDefault: false
|
|
editable: false
|
|
```
|
|
|
|
---
|
|
|
|
## 5. Migration
|
|
|
|
### 5.1 Migrationspfad
|
|
|
|
```
|
|
Phase 1: VictoriaMetrics Deployment (Zero Downtime)
|
|
═══════════════════════════════════════════════════
|
|
1. VictoriaMetrics Container hinzufügen
|
|
2. Parallel zu Prometheus laufen lassen
|
|
3. Grafana Datasource auf VM umstellen
|
|
4. Prometheus Container entfernen
|
|
|
|
Phase 2: DuckDB Integration
|
|
═══════════════════════════
|
|
1. DuckDB Dependency hinzufügen
|
|
2. Analytics Service implementieren
|
|
3. Cron-Job aktivieren
|
|
4. API Endpoints erstellen
|
|
5. Grafana Dashboard für Business Metrics
|
|
|
|
Phase 3: Historische Daten (Optional)
|
|
═════════════════════════════════════
|
|
1. Prometheus Daten exportieren
|
|
2. In VictoriaMetrics importieren
|
|
3. Initiale DuckDB-Befüllung aus Prometheus
|
|
```
|
|
|
|
### 5.2 Rollback-Plan
|
|
|
|
**VictoriaMetrics → Prometheus:**
|
|
- Gleiche Config-Datei funktioniert
|
|
- Grafana Datasource URL zurückändern
|
|
- Container tauschen
|
|
|
|
**DuckDB:**
|
|
- Service deaktivieren
|
|
- Keine Abhängigkeiten in anderen Services
|
|
|
|
---
|
|
|
|
## 6. Monitoring & Alerting
|
|
|
|
### 6.1 VictoriaMetrics Self-Monitoring
|
|
|
|
```yaml
|
|
# prometheus/alerts.yml (funktioniert auch mit VM)
|
|
groups:
|
|
- name: victoriametrics
|
|
rules:
|
|
- alert: VMStorageSpaceLow
|
|
expr: vm_free_disk_space_bytes / vm_available_disk_space_bytes < 0.1
|
|
for: 5m
|
|
labels:
|
|
severity: warning
|
|
annotations:
|
|
summary: "VictoriaMetrics disk space low"
|
|
```
|
|
|
|
### 6.2 DuckDB Health Check
|
|
|
|
```typescript
|
|
// Endpoint: GET /api/analytics/health
|
|
{
|
|
"status": "healthy",
|
|
"database_size_bytes": 10485760,
|
|
"last_snapshot": "2025-01-28",
|
|
"total_records": 365
|
|
}
|
|
```
|
|
|
|
---
|
|
|
|
## 7. Backup-Strategie
|
|
|
|
### 7.1 VictoriaMetrics
|
|
|
|
```bash
|
|
# Snapshot erstellen (built-in)
|
|
curl -X POST "http://victoriametrics:8428/snapshot/create"
|
|
|
|
# Backup zu S3/MinIO
|
|
vmbackup -storageDataPath=/storage -snapshot.createURL="http://localhost:8428/snapshot/create" -dst=s3://backups/vm/
|
|
```
|
|
|
|
### 7.2 DuckDB
|
|
|
|
```bash
|
|
# Einfacher File-Copy (konsistent da single-writer)
|
|
cp /data/metrics.duckdb /backup/metrics-$(date +%Y-%m-%d).duckdb
|
|
|
|
# Oder Parquet-Export für Archivierung
|
|
duckdb /data/metrics.duckdb -c "COPY daily_metrics TO '/backup/metrics.parquet' (FORMAT PARQUET)"
|
|
```
|
|
|
|
---
|
|
|
|
## 8. Kosten & Ressourcen
|
|
|
|
### 8.1 Storage-Projektion (2 Jahre)
|
|
|
|
| Komponente | Jetzt | Nach Migration |
|
|
|------------|-------|----------------|
|
|
| Prometheus | 5 GB (30d) | 0 GB (entfernt) |
|
|
| VictoriaMetrics | 0 GB | ~15 GB (2y) |
|
|
| DuckDB | 0 GB | ~20 MB (2y) |
|
|
| **Total** | **5 GB** | **~15 GB** |
|
|
|
|
### 8.2 RAM-Projektion
|
|
|
|
| Komponente | Jetzt | Nach Migration |
|
|
|------------|-------|----------------|
|
|
| Prometheus | ~2 GB | 0 GB |
|
|
| VictoriaMetrics | 0 GB | ~500 MB |
|
|
| DuckDB | 0 GB | ~50 MB (on-demand) |
|
|
| **Total** | **~2 GB** | **~550 MB** |
|
|
|
|
---
|
|
|
|
## 9. Implementierungsplan
|
|
|
|
### Phase 1: VictoriaMetrics (Tag 1)
|
|
- [ ] docker-compose.macmini.yml aktualisieren
|
|
- [ ] VictoriaMetrics Container hinzufügen
|
|
- [ ] Grafana Datasource konfigurieren
|
|
- [ ] Bestehende Dashboards testen
|
|
- [ ] Prometheus Container entfernen
|
|
|
|
### Phase 2: DuckDB Service (Tag 1-2)
|
|
- [ ] duckdb Package installieren
|
|
- [ ] AnalyticsModule erstellen
|
|
- [ ] DuckDB Schema initialisieren
|
|
- [ ] Daily Snapshot Cron-Job
|
|
- [ ] API Endpoints für Grafana
|
|
|
|
### Phase 3: Dashboards & Dokumentation (Tag 2)
|
|
- [ ] Business Metrics Dashboard erstellen
|
|
- [ ] Master Overview Dashboard aktualisieren
|
|
- [ ] Dokumentation finalisieren
|
|
- [ ] Backup-Scripts erstellen
|
|
|
|
---
|
|
|
|
## 10. Entscheidungsmatrix
|
|
|
|
| Kriterium | Gewicht | Prometheus | VM + DuckDB | Score |
|
|
|-----------|---------|------------|-------------|-------|
|
|
| Langzeit-Retention | 30% | 2/10 | 10/10 | +2.4 |
|
|
| Ressourceneffizienz | 20% | 4/10 | 9/10 | +1.0 |
|
|
| Migrationsaufwand | 15% | 10/10 | 8/10 | -0.3 |
|
|
| Wartbarkeit | 15% | 7/10 | 8/10 | +0.15 |
|
|
| Analytics-Fähigkeit | 10% | 3/10 | 9/10 | +0.6 |
|
|
| Backup-Einfachheit | 10% | 5/10 | 9/10 | +0.4 |
|
|
| **Gesamt** | 100% | **4.7/10** | **9.1/10** | **+4.4** |
|
|
|
|
---
|
|
|
|
## 11. Risiken & Mitigationen
|
|
|
|
| Risiko | Wahrscheinlichkeit | Impact | Mitigation |
|
|
|--------|-------------------|--------|------------|
|
|
| VM nicht 100% PromQL-kompatibel | Niedrig | Mittel | Dashboards vorab testen |
|
|
| DuckDB Datenverlust | Niedrig | Hoch | Tägliches Backup |
|
|
| Cron-Job Ausfall | Mittel | Niedrig | Monitoring + Catch-up Logic |
|
|
| Storage voll | Niedrig | Mittel | Alerting bei 80% |
|
|
|
|
---
|
|
|
|
## 12. Referenzen
|
|
|
|
- [VictoriaMetrics Dokumentation](https://docs.victoriametrics.com/)
|
|
- [VictoriaMetrics vs Prometheus Benchmark](https://valyala.medium.com/prometheus-vs-victoriametrics-benchmark-on-node-exporter-metrics-4ca29c75590f)
|
|
- [DuckDB Dokumentation](https://duckdb.org/docs/)
|
|
- [Grafana JSON Datasource](https://grafana.com/grafana/plugins/simpod-json-datasource/)
|
|
|
|
---
|
|
|
|
## Appendix A: Bestehende Dashboards
|
|
|
|
| Dashboard | UID | Änderung nötig |
|
|
|-----------|-----|----------------|
|
|
| System Overview | `system-overview` | Keine |
|
|
| Backends & Docker | `backends-docker` | Keine |
|
|
| Application Details | `application-details` | Keine |
|
|
| Database Details | `database-details` | Keine |
|
|
| User Statistics | `user-statistics` | Keine |
|
|
| Master Overview | `master-overview` | Business Metrics hinzufügen |
|
|
|
|
## Appendix B: Prometheus Config Kompatibilität
|
|
|
|
Die bestehende `prometheus.yml` funktioniert ohne Änderung mit VictoriaMetrics:
|
|
|
|
```yaml
|
|
# Alle Scrape-Configs bleiben identisch
|
|
scrape_configs:
|
|
- job_name: 'mana-core-auth'
|
|
static_configs:
|
|
- targets: ['mana-core-auth:3001']
|
|
metrics_path: '/metrics'
|
|
scrape_interval: 30s
|
|
# ... alle anderen Jobs
|
|
```
|