Brain overloaded — slow dashboard, hub rebooting

Too many open dashboards can saturate the brain and starve the hub link; how to see it in the logs, and what to do.

Many browsers (or one browser with many tabs) polling the computed fleet view can saturate the brain. The hub's API session then starves, the hub link flaps, and in the worst case the hub's own watchdog reboots it. From the dashboard it looks like HUB LINK DOWN coming and going while everything is powered.

docker logs --tail 300 dsc-hub-brain

Look for request bursts on /fleet/computed, session restart lines (session looks dead — restarting (#n, next no sooner than … s); the wait doubles from 30 s up to 8 minutes), and hub reconnects. Read the logs before theorising; a flapping link has other causes (Wi-Fi roaming).

1. Close spare dashboards and tabs; a wall display counts.
2. Raise the Camera thumbnail refresh interval on phones.
3. Settings › Devices › Poll rates: leave the defaults unless the logs show ingest itself is the load.
4. If the brain is wedged: docker stop -t 20 dsc-hub-brain then docker start dsc-hub-brain. Not restart, which has hung the Pi.

A hub uptime that does not move across a brain restart is the estimate carrying between polls, not a boot loop. Watch HB # instead.

Guide contents  ·  Troubleshooting  ·  Troubleshooting  ·  Digital Stealth Care

DSC-Brain user guide · Brain overloaded — slow dashboard, hub rebooting · this page is edited on the site; the Notion original was retired after the mirror