NMC shut down 1 of 4 servers for no apparent reason

[JRVcr]

2 x Smart-UPS SRT 2200 with NMC3 and temperature probe set up as redundant UPSs.

"High" temp is set to 24 degrees C, "Maximum" to 27 degrees C.

NMC Logs show server room got to 24 degrees at 8:24 AM when the door was left open. Temperature never reached 27 degrees; after 8 minutes it dropped back below the 24 degree threshold.

Control Policy on both NMCs is "Environment: A critical fault exists for integrated Environmental Monitor input contact ({name} at {location}): Graceful shutdown."

Here's what's in the PCNS log from the server that was shut down:

1. Why did it send a shutdown when only "High" temperature was reached, not "Maximum", since policy specifies "critical" event?
2. Why did it send the shutdown to only 1 connected server? All show connectivity to both UPS's. The other servers show the temp out-of-range events but no shutdown events. I'm glad they didn't, here; the one shutdown caused enough chaos! But I DO want the other servers to shut down if it's too hot. (27 degrees, though not 24.)

[BillP]

@JRVcr 

1. Why did it send a shutdown when only "High" temperature was reached, not "Maximum", since policy specifies a "critical" event? PowerChute will trigger the shutdown when the first threshold has been reached. If you do not want the shutdown triggered for the high-temperature violation, deselect that within the NMC or set the high-temperature threshold to 27. 
2. You should verify you have configured PowerChute on the other server to trigger a shutdown for a temperature threshold violation. 

[JRVcr]

Bill, you're correct that the one server that shut down has "Temperature Probe 1 Out Of Range" enabled for shutdown in PCNS and the others do not. So I need to fix that. 

But, back in NMC, when I add a Control Policy, I'm offered the choice of "Warning" or "Critical".

I chose "Critical."

I had expected the NMC would only signal shutdown if the "Critical" threshold was passed. Not the "Warning" threshold. That's not correct? If so, what's the point of having Policies configurable for the two thresholds?

The goal was to have the warning at 75 F so one of the people who receives the notifications can block the door open and set up a box fan, where shutdown can be averted by keeping the temp below 80 F until air-conditioning is restored.

[BillP]

@JRVcr 

PowerChute does not differentiate. It will react to either as a critical event. The solution is to add a second AP9335 to port two. Configure port two with the high temp threshold that will send a notification.

PowerChtue will only react to port one so configure port one with the max temp threshold that will command PowerChute off.

[JRVcr]

@BillP , thanks for your reply!

If I leave the NMC configured as it is--to send a shutdown signal at 27 C--and disable Shutdown on the temperature violation Event in PCNS, will the NMC shut the servers down at 27 C? In other words, is the root of my issue that--at least on this one server--over-temperature shutdown is configured in both NMC and PCNS, when, given my objectives (notification at 24 C, shutdown at 27 C), it should only be configured on the NMCs?

Or does PCNS ignore the shutdown signal if it hasn't also been enabled in the PCNS Event?

[BillP]

@JRVcr 

Are you using an AP9335 with the AP9810? If so you will need to create two Control Policies. One for the AP9810 and one for AP9335T/TH

On the AP9810 jumper 

Zone A Com to Output Com

Zone A NC to Output NC

When the Max temp threshold is breached, the IO state changes, and the UPS powers off and signals PowerChute. PowerChute will shutdown because of the critical UPS turn off event.

PowerChute should not be configured to shutdown for a temperature threshold violation and should not be configured to power off the UPS or Outlet Group. 

 

[JRVcr]

I won't be at this client again until Monday, so SKUs are hard to come by at the moment. But we are NOT using AP9810s. The thermal sensors connect directly to the NMC3s.