What collector schedules / frequencies do you use?

[thecodygriffin]

I am curious how frequently others are scheduling their collectors to send data to the API.

I am using a hub-spoke model with the web/API and database deployed with Docker and the metrics and Btrfs binaries deployed on various bare metal machines using cron for scheduling.

I've scheduled cron to send data every 15 minutes for both the metrics and Btrfs binaries because it was the example provided in the https://github.com/Starosdev/scrutiny/blob/7d298d2fdc7cab2d589c179d1681c8894726263d/docs/INSTALL_HUB_SPOKE.md doc. It feels like overkill, but I am honestly not sure.

What collectors do you use and with what schedules?

[Starosdev]

The */15 schedule in docs/INSTALL_HUB_SPOKE.md is an example, not a recommendation that everyone should follow.

For the core SMART collector, */15 is usually more frequent than most installations need. The historical metrics pipeline was originally designed around much lower collection frequency, and the InfluxDB troubleshooting docs explicitly call out daily collection as the expected baseline for SMART trend data. Running every 15 minutes increases the number of stored and rendered points substantially without adding much value for normal disk health monitoring.

Maintainer-style guidance would be:

• SMART health monitoring: start with daily
• If you want somewhat faster visibility into changes: hourly or every 6-12 hours
• Use */15 only when you have a specific reason to want high-resolution trend data, or while actively testing/troubleshooting

If you are also running other collectors like filesystem, ZFS, Btrfs, or performance-related collection, those can justify a shorter interval depending on what you are trying to observe. But for plain SMART collection in a hub/spoke setup, daily or hourly is usually the better default.

So yes: if 15 minutes feels like overkill for SMART, that is a reasonable conclusion.