higher cpu usage on the nodes that pulse monitors

[jodewee]

Pulse 4.14.0

I notice that once i added a proxmox or PBS node to pulse,
the cpu usage on that specific node goes consistent higher.
Also the power consumption (probably related to the cpu usage)

maybe it is related to polling frequency?
Is this known as is, or is there maybe a tweak?

[samwathegreat]

@jodewee it seems reasonable that this would happen due to the constant polling of the proxmox API. Not sure this is a bug, although it could likely be optimized...

[rcourtman]

this is expected behavior and not a bug. pulse polls the proxmox API every 10 seconds to provide near-real-time monitoring data. each poll cycle makes multiple API requests to gather node status, VM metrics, storage info, etc.

why this causes higher cpu usage:

• the proxmox API server (pveproxy) processes these requests
• more frequent polling = more CPU cycles processing API requests
• this is normal for any monitoring tool

comparison with other tools:

• prometheus/grafana: typically 15-60 seconds (configurable)
• zabbix: 30-60 seconds (configurable)
• nagios: 60-300 seconds (configurable)
• pulse: 10 seconds (hardcoded)

current limitation:
the 10-second polling interval for PVE nodes is currently hardcoded and not user-configurable (PBS polling interval can be configured). see internal/api/config_handlers.go:2099

potential improvements:

1. make PVE polling interval configurable (similar to PBS)
2. implement smart polling (slower polling when no changes detected)
3. reduce number of API calls per cycle

this would be a feature request rather than a bug fix. the current behavior is by design to provide responsive real-time monitoring

[baegopooh]

Quote from readme.

"Quick start - most settings are in the web UI:

Settings → System: Polling intervals, timeouts, update settings"

I was looking for this option to configure polling interval everywhere until..... stumbling here.

Please add this feature or at least delete that line from readme

[tranb3r]

@rcourtman
With 4.32.1, I have set "Monitoring cadence" to 5 minutes.
However, pulse-sensor-proxy is still logging "temperature data fetched successfully" every 10 seconds.
Is this normal ?