Raise an (Alert) Event for a group of sensors

[kt-alt]

A few days ago, my server unexpectedly shutdown due to high temp. I've installed glances monitor it and can see the temp readings, but it doesn't appear that an alert or event is raised for temperature.

I have previously modified event.py to create syslog messages for events, which are sent over to graylog for centralized alerting. I'd like to do this for my server temp as well.

Is there a way to enable this? Or is this something to suggest as a new feature.

[nicolargo]

Hi @kt-alt

yes, you can configure thresholds for the sensors in the Glances configuration file:

[sensors]
# Documentation: https://glances.readthedocs.io/en/latest/aoa/sensors.html
disable=False
# By default refresh every refresh time * 2
#refresh=6
# Hide some sensors
#hide=ambient
# Sensors core thresholds (in Celsius...)
# Default values are grabbed from the system
temperature_core_careful=60
temperature_core_warning=70
temperature_core_critical=80
...

[kt-alt]

Thanks, I misunderstood the note at the bottom of this part of the docs - which noted no alerts on sensors.

https://glances.readthedocs.io/en/latest/aoa/sensors.html

I did try to stress the system to trigger a temperature warning but was unsuccessful. Perhaps I didn't exceed the default values from the system, so will try setting the custom values.

[kt-alt]

@nicolargo - I set some purposefully low custom thresholds in the conf file and was able to see the the temperature readings in the UI were red (meaning the thresholds were adjusted). However they did not trigger any events either in the UI or in the syslog.

Is this something possible or could be done with small code adjustments? It wasn't obvious from the event.py what customizations might be needed and I'd rather not break stuff blindly.

Thanks.

[vlanx]

Did you manage to pull this off? I'm interested in the same setup, but even changing the threshold values to some low ones, doesn't change the color output in the dashboard.

@nicolargo Do you have any idea why the threshold values are not being taken into account?

[kt-alt]

Not fully, I eventually gave up and switched to prometheus as I think I was trying to get glances do stuff too outside of what it was intended to do.

[nicolargo]

I look deeper in the code and the current behavor is the following:

• search limit define for a specific sensor in the Glances configuration file (ex: temperature_core_Core 4_careful=10, temperature_core_Core 4_warning=20,temperature_core_Core 4_critical=30)
• if the following limit is not define, the uses system one

Perhaps a enhancement will be:

• search limit define for a specific sensor in the Glances configuration file (ex: temperature_core_Core 4_careful=10, temperature_core_Core 4_warning=20,temperature_core_Core 4_critical=30)
• if the previous limit is not define, search limit define for a type sensor in the Glances configuration file (ex: temperature_core_careful=10, temperature_core_warning=20,temperature_core_critical=30)
• if the following limit is not define, the uses system one

What do you think ?

[vlanx]

Hi @nicolargo,

I do agree that the enhancement you proposed would give even more granularity in defining the warning and critical values for the sensors, which, in my opinion, is a good thing. For example, I would be interested in defining the temperatures for all cores at once, I would then use temperature_core_warning=40.

But even now, defining the various temperature limits is still not possible for me. Glances is not taking them into account.

# glances --version
Glances version:        4.2.1
Glances API version:    4
PsUtil version:         6.1.0
Log file:               /root/.local/share/glances/glances.log

[nicolargo]

Merged into develop (see #3049)

@kt-alt @dot-1q can you test the feature ?

[vlanx]

Hi @nicolargo,
Thank you for your quickness in viewing this issue and providing help and a solution!

Today I tested your dev branch locally and the colors are now being outputted correctly on my machine.
I can set colors both globally for the temperature sensor, and also for a specific one (Core 0). I tried to showcase that in the screenshot.

It is working as I would've expected now. Thank you so much again!