Juniper PEM Errors – Difference Between ‘Not OK’ and ‘Input Failure’

juniperrouter

I have one query with regard to PEM (Power Entity Modules) in Juniper routers. On MX series routers, 4 x PEM Modules are installed.

However in case 1 x PEM is not receiving power (like Input Power Supply is gone ) but PEM itself is physically there then Juniper router reports 2 x chassis alarms , with one being that 'PEM Not Ok' while other 'PEM Input Failure'.

Below is example that I took from Juniper KB (https://kb.juniper.net/InfoCenter/index?page=content&id=KB29614&cat=SRX_5400&actp=LIST)

Only PEMs 0 and 1 are cabled. The result is these alarms:

lab@mx240> show chassis alarms 
4 alarms currently active
Alarm time               Class  Description
2014-10-23 17:43:35 UTC  Major  PEM 3 Input Failure
2014-10-23 17:43:35 UTC  Major  PEM 3 Not OK
2014-10-23 17:43:35 UTC  Major  PEM 2 Input Failure
2014-10-23 17:43:35 UTC  Major  PEM 2 Not OK

In the KB they mentioned the CAUSE like this :-

This may be a hardware issue with the PEM, or an issue with the power supply to the PEM.

and solution they mentioned is like this :-
"If a PEM was cabled, but the circuit breaker was shut off, or the PEM was not receiving any voltage for some other reason, the same alarms and show chassis power output would be seen. It is important to verify that the PEM is receiving the correct voltage (which may require an electrician) before assuming it is a hardware issue with the PEM"

How could we distinguish between two cases? How could we know that whats the real fault? Is it PEM become faulty (PEM Not Ok) or is it input to PEM gone?

We have to extend the such alarms to both Network Field Team and Power Team and hence this distinguish is necessary. In case PEM Input Not OK , the power team will take action and would fix the source. and in case PEM went faulty, Network field team guys will inspect and replace PEM Module.

Thanks

Best Answer

First and foremost "Not OK" is just signifying that there is some kind of problem relating to the PEM, it could be anything. The "Input Failure" is just trying to be more detailed (this doesn't always happen).

Unfortunately, for these sorts of alarms there is no way to definitively determine if the cause is due to the PEM in the router or some problem with the power source from looking at the router.

It's really unlikely that 2 PEMs are failing at the same time. Most power in Central Offices and Data Centers will be connected to an A side and a B side PDUs for redundancy. For example, PEM 0/Feed 1 and PEM 1/Feed 1 will connect to A, with PEM 0/Feed 2 and PEM 1/Feed 2 connecting to B.

Looking at your situation, my first instinct would be to have someone look at the PDUs to see if a breaker is tripped or turned off.