Trouble Creating a Certain Alert

Hi All,

I have a bit of a brain boggler and was wondering if someone could lend me a hand.

The alert in question monitors the Air Conditioning in a small server room. There are two AC units in the server room, one is on standby and the other is active. These swap roles each weekend.

Due to an incident that occurred a few months ago I have set up custom pollers on these two nodes to monitor Fan speed. The standby is at 0 and the active is around 600 (measured in RMP).

I want to be alerted if both of the customer pollers report back as 0 but I cannot get Orion to see the two customer pollers as one entity in this instance.

Does anyone have anything that could help me?

Thanks

Chris

Find more posts tagged with

intelligent alerting

active alert

custom pollers

air conditioning

Alerting

Accepted answers

ecfrs

Hi All,

Thanks for all your help. I have managed to get this working:

All comments

d09h

Can the standby/ active status change generate a trap? Which MIB are you polling? Which vendor/ model?

jhandberg

What about a modification of Tip #3 from this page?

https://thwack.solarwinds.com/community/solarwinds-community/product-blog/blog/2015/03/12/cutting-down-on-alerting-noise…

It seems like a complex alert with 2 parts, one for each unit might do the job. Just modify it do work with your 2 custom pollers.

blsanner

Another idea might be to create a UnDP transform that adds the two UnDPs for fan speed together. Then alert when the transform value is zero meaning both of the UnDPs have a value of zero.

d09h

Can that be done across two nodes?

blsanner

It just hit me that you can't do that and I came back here to reply. No, the best solution will be what jhandberg mentioned and creating a complex alert condition. That's what I get for posting too early in the morning.

d09h

If trap exists for this, that would be ideal. If ecfrs can supply vendor SNMP info or can SNMPwalk the device, it may be that there is a MIB for the standby/ active role.

ecfrs

Hi All,

Thanks for the response. Great to see an actual community for a product!!

So, the issue I had was that the AC units turned off, when they came back online both went to standby (one is usually in standby and the other is active).

The vendor is Airdale but I cannot see any Trap options on the devices which is a pain.

I imported a load of UDP's and the one that changes that I'm trying to alert on is Airdale_Evaporator_Fan_Speed (MIB Name: Vendor-MIB:emergingTech OID 1.3.6.1.4.1.9839.2.1.2.13.0)

The standby node is at 0 and the active shows a figure but not one that is static as it depends on how fast the fan is moving to meet the temp we need the room at. I couldn't see an option to put in status for standby/active although I am unsure if they both read as standby when the issue occurred (tried to recreate the issue but have been unsuccessful). The only thing I know is that both fans had stopped and we only got alerted when the environment sensors recorded high temp and by that time (as it was out of hours) it was too late.

Current standby Node

Current Active Node

I have looked into the Advanced alert options as the description of it looks like its what I need but I'm sure that I'm not thinking the way the alert wants me to think (the 2 objects are the custom pollers for each node)

jhandberg

This looks reasonable.

ecfrs

Yeah, it looks like the most logical way to set up this alert. I've also tried setting up the scope of the alert to be the custom pollers instead of the actual nodes but still no luck. Has anyone got any ideas? I've read the blog that John Handberg posted previously and feel that I've done what is necessary. So confused, usually so easy to set up alerts, however complex.

m_roberts

If it was me, I would be inclined to create two Group entries within the trigger condition.

Top level group - AND

Sub Group - AND

Custom Node Poller Instance Name = <instance 1>

Custom Node Poller Current Value = 0

Sub Group - AND

Custom Node Poller Instance Name = <instance 2>

Custom Node Poller Current Value = 0

Easier to visualise.

Also, as you have a defined Instance name is equal to statement, you do not need to populate the Scope of Alert section with the node names.

Mark

ecfrs

Hi All,

Thanks for all your help. I have managed to get this working: