Setup for Python 3.x to Query LM looking at Arista Voltage Sensors - Just PowerSupply

I'm not sure how Python/LM API and some of the data not showing up are related. Are you looking to use the LM API to figure out where these situations exist?

Stuart,  Yes I am looking to use the LM API to cycle through the configuration and find these devices that need further insight.

Else I look at these by hand by looking for Arista Voltage Sensor - and then looking through the Search Results list one by one. 
Making a note about each device that has an issue so we can do follow up with the Facility the device resides in and verify the Power Cords, etc. 

Ok, in that case, I recommend the Python SDK. Documentation and examples.

Will do - - I have looked at that in the past, but I was looking for more examples like you have in your tag - links.

I did notice that something changed in your tag link for the first link - >>My personal library of LogicModules is here.<<  
For me, the tag 'here'  did not find a page. 
Your other links worked though.

Thanks for the pointers - I'll review and start building my library of examples.

Yeah, my repo had to come down.

FWIW: why not just set a threshold of 0 on that datapoint? If you already have a threshold, create a complex datapoint that uses that datapoint as its expression. Then you can set a specific threshold of 0 on that datapoint.

The standard way to put an alert on this does not appear to be able to only put this on the PowerSupply items, but all the Volt reporting widgets for Arista.
That is not what we wanted to have happen as that appears to also be setting the POE Voltage settings as well. 

If I would limit it to just the PowerSupply item that would be great.  We have 133 of these devices at present and still refreshing the enterprise. So more are coming.  Setting all of these by a manual process is not the desired way to do this alerting.  

Now that I've been looking at them by hand, I am also noticing that some are in the 220 Volt area and some are in the 110 Volt area.  Surprise not all of our facilities have 220 as a source of power. 

I would love it if this could be enabled for alerting on low power on the PowerSupply widget once a device is created / discovered. 

Ok, i'm getting a better picture of your problem (not clear but getting there). You need the threshold to apply to some instances but not others. And that means setting the threshold manually on each instance, which is not scalable.

The traditional way of handling this is to split the datasource into two copies: one for the instances with one threshold and one for instances with the other. You limit the discovery of each datasource to include only the instances in that group (discovery filters are easiest). That way you have one datasource that is doing what it's doing now, except not monitoring the instances that need the different threshold. The new datasource is only monitoring the instances that need the new threshold. This lets you set the threshold globally, but have it only apply to certain objects.

Is there a way to programmatically (i.e. querying the device during discovery) determine which ones need the other threshold?

In answer to the 'is there a way to programmaticly' find that.

I'm not sure - - an example is for one of the Arista devices - it has an item in
Arista_Voltage_Sensors -> PowerSupply1 (or 2) 

Within that is then the Voltage Sensor Overiew which has the Input Voltage Sensor (100711104)   Volts and Status - -I want Volts.

I would assume it is in the MIB - but I have not explored to find the actual MIB OID yet. 

When I click to look at the DataSources -
That takes me to the DataSource - Arista_Sensor_Voltage,
it looks like it is in the OID_MAP .1.3.6.1.2.1.47.1.1.1.1.2  for the description
and  .1.3.6.1.2.1.47.1.1.1.1.4  for the Contained_in

I used to use GetIF for the SNMPWALK - and to use the MIB OID mapping to explore Tables. 

As an Aside - why does the oid_map have mappings that for some start with '.1.3.6.1.'  and yet the 'description' entry does not start with '.1'  but with '1.' ? 
If this is really an OID - I would expect to see them all have the same form -  all start with a '.'  and then follow the OID chain to the actual value.

The leading dot is optional in OID addresses. It used to be required by RFC1213 standards IIRC, but newer software libraries have accounted for people leaving off the dot. Ours is also one that tolerates omitting the leading dot. This is otherwise known as a "fortunate typo".

Let's lay down some terminology so we make sure we're all talking about the same things:

Arista_Voltage_Sensors <- this is the name of the datasource.

Arista Voltage Sensors <- this is the display name of the datasource. It shows up directly under the device in the resource tree.

PowerSupply1 (or 2) <- this is the instance group. In this case, the group is created based on the value of auto.sensor_owner on the instance. 

Input Voltage Sensor (100711104) <- this is the instance, you'll have one instance per sensor.

Voltage Sensor Overview <- this is an "overview" graph, meaning that it shows data from all sensors on the device. In this case, one line per instance.

Volts <- this is technically the overview graph datapoint name, which in this case is the same as the datasource's datapoint name.

As I understand your ask: you want to put a custom threshold on the volts datapoint, but only have that threshold apply to instances where the instance group starts with "PowerSupply". There may be other sensors, but they may not be in an instance group called "PowerSupply" and you don't want those sensors to generate an alert when the voltage is 0. Is that correct?

You were able to re-state the problem perfectly - - Yes- I am only looking for those instances that start with PowerSupply  -