Optimize scrape performance and resource usage
mainScrape performance varies significantly by application. Use these guidelines to tune your scrape_interval and container limits:
Scrape Timing Guidelines
- Bazarr: Can take tens of seconds (especially with episode metrics). Set
scrape_intervalto60sor more. - Sonarr/Lidarr: Takes a few seconds due to per-series/artist fan-out.
- Radarr: Takes 1-2 seconds (large JSON payload).
- SABnzbd/Prowlarr: Very fast.
Important: If a scrape arrives while the previous one is still running, Exportarr skips it and raises an error gauge. Always set your scrape_interval longer than your slowest expected scrape.
Resource Management
- Memory: Expect 25–100 MB RSS. High-end usage occurs during Bazarr episode walks or large Radarr movie lists. In Kubernetes, set
GOMEMLIMITto the container memory limit to manage GC spikes. - Slow Scrapes: If scrapes are too slow, do not simply decrease
REQUEST_TIMEOUT. Instead, useDISABLE_*flags (e.g.,DISABLE_HISTORY_METRICS,DISABLE_WANTED_METRICS,DISABLE_EPISODE_METRICS) to remove the expensive endpoints entirely.