Scraping Multiple Databases

The scraper can collect from many Oracle databases in the same process. Each database entry has a local name, and every PostgreSQL row includes that name in source_database.

Grafana dashboards normally select one source database at a time. This keeps high-cardinality views such as DAH, SQL performance, and current sessions usable when many databases are collected into the same PostgreSQL backend.

If you are connecting to multiple databases with mTLS, see mTLS for multiple databases with Oracle Wallet.

Example

databases:
  db1:
    username: ${DB1_USERNAME}
    password: ${DB1_PASSWORD}
    url: localhost:1521/freepdb1
    queryTimeout: 5
    connMaxLifetime: 30m
    connMaxIdleTime: 10m
    maxOpenConns: 10
    maxIdleConns: 10

  db2:
    username: ${DB2_USERNAME}
    password: ${DB2_PASSWORD}
    url: localhost:1522/freepdb1
    queryTimeout: 5
    connMaxLifetime: 30m
    connMaxIdleTime: 10m
    maxOpenConns: 10
    maxIdleConns: 10

metrics:
  scrapeInterval: 15s

operational:
  enabled: true
  interval: 1m
  queryTimeout: 10s

performance:
  sqlPlans:
    enabled: true
    interval: 2m
    topN: 20
    queryTimeout: 10s

output:
  postgresql:
    url: ${POSTGRES_URL}
    autoMigrate: true
    retention: 720h

Use one entry per service or PDB that must be monitored independently. For RAC databases, point the connection string at the service you want to monitor; the queries use GV$ views where instance-level data is needed.

Per-Database Metric Selection

Additional TOML/YAML metrics run against every configured database unless the definition has a databases field.

[[metric]]
context = "db_platform"
labels = [ "platform_name" ]
metricsdesc = { value = "Database platform." }
request = '''
select platform_name, 1 as value
from gv$database
'''
databases = [ "db2", "db3" ]

The example above only runs for the database entries named db2 and db3. If databases is omitted, the metric runs for all configured databases. If databases = [], the metric is disabled.

The built-in performance collectors for DAH, SQL performance, current sessions, and blocking sessions are database-wide collectors. They write to dedicated PostgreSQL tables and rely on dashboard variables for database selection.

Duplicate Connections

If more than one database entry uses the same URL and username, the scraper logs a warning similar to:

msg="duplicated database connections" "database connections"="db1, db2 count=2

Duplicate entries are allowed, but they usually create duplicated PostgreSQL samples unless each entry has an intentional purpose.