GraphQL metrics temperature query spins up all array disks

Environment

Unraid OS Version: 7.3 (Unraid OS)

Are you using a reverse proxy? No

Pre-submission Checklist

• I have verified that my Unraid OS is up to date
• I have tested this issue by accessing my server directly (not through a reverse proxy)
• This is not an Unraid Connect related issue (if it is, please submit via the support form instead)

Issue Description

Calling the metrics GraphQL query to fetch temperature data triggers a spin-up of all array disks. The first request after disks have spun down takes ~5 seconds to respond because it waits for every disk to spin up. Subsequent requests are fast since the disks are already active.

This affects any application that polls temperature data periodically (e.g., monitoring dashboards), as each poll cycle causes unnecessary disk spin-ups, increasing wear, power consumption, and noise.

Steps to Reproduce

1. Ensure all array disks are in a spindown (sleep) state.
2. Execute the following query against the GraphQL API:
   query {
   metrics {
   temperature {
   summary {
   average
   hottest { name, current { value, status } }
   }
   }
   }
   }
3. Observe that the response takes several seconds and all array disks spin up during the request.

Expected Behavior

Temperature data comes from hardware sensors (e.g., lm-sensors, /sys/class/thermal/) and should be completely independent of array disk I/O. The query should return quickly regardless of whether disks are spun down.

Actual Behavior

All array disks are woken up during the temperature query. The first request after spin-down takes approximately 5 seconds to complete, as it waits for every disk to spin up before returning a response.

Additional Context

Questions:

1. Does the temperature endpoint iterate over array disks for some reason? Could it use a sensor-only path that
   doesn't trigger disk access?
2. Is there a way to fetch temperature data without waking disks (e.g., reading from a cached value or a daemon that polls sensors independently)?

Start by tracing the GraphQL metrics temperature query and its temperature endpoint, then reproduce the request with array disks spun down. Check which temperature data path causes disk access; done means the query returns without waking array disks and still reports the expected temperature summary.