Author Topic: How to Alert on Log entry "CPU overloaded"  (Read 456 times)

Offline gniknalu

  • Full Member
  • ***
  • Posts: 148
  • Karma: +3/-2
How to Alert on Log entry "CPU overloaded"
« on: April 04, 2018, 12:12:20 pm »
When the Very Secure crashed, according to the support team, they saw "CPU overloaded" events in the log. [ReportError nm_cpu / CPU overloaded, count: 5, sec: 1206, idle: 3, system: 40, delay: 184 / <0?771df000>] I'm wondering if there is a way to alert when something similar appears in the log and maybe even trigger a reboot?

Thank you

Online akbooer

  • Master Member
  • *******
  • Posts: 5752
  • Karma: +246/-69
  • "Less is more"
Re: How to Alert on Log entry "CPU overloaded"
« Reply #1 on: April 04, 2018, 01:35:11 pm »
Not directly related to the log entry, but several plugins (EventWatcher, SystemMonitor?) expose a CPU-load type of parameter.

In particular, EventWatcher has a CPU05 variable giving CPU loading over a 5-minute average.

If really necessary, you could pull this directly from the pseudo-file in /proc/loadavg, without even needing a plugin installed.

_________________

Edit: I should add that I routinely log memory usage (free and available) plus CPU load for all three of my Vera systems. See attachment
« Last Edit: April 04, 2018, 01:39:32 pm by akbooer »
3x Vera Lite-UI5/Edge-UI7, 25x Fibaro, 23x TKB, 9x MiniMote, 2x NorthQ Power, 2x Netatmo, 1x Foscam FI9831P, 2x Philips Hue
Razberry, MySensors Arduino, HomeWave, AltUI, DataYours, openLuup, ZWay, ZeroBrane Studio.

Offline gniknalu

  • Full Member
  • ***
  • Posts: 148
  • Karma: +3/-2
Re: How to Alert on Log entry "CPU overloaded"
« Reply #2 on: April 11, 2018, 11:25:31 am »
Not directly related to the log entry, but several plugins (EventWatcher, SystemMonitor?) expose a CPU-load type of parameter.

In particular, EventWatcher has a CPU05 variable giving CPU loading over a 5-minute average.

If really necessary, you could pull this directly from the pseudo-file in /proc/loadavg, without even needing a plugin installed.

Edit: I should add that I routinely log memory usage (free and available) plus CPU load for all three of my Vera systems. See attachment

Yeah - but none of the ones mentioned alert on problems - just log them. To me it makes sense that there should be some alert when an event occurs in the log that is indicating instability or a possible crash. . . .