A possible failure for MX and QFX devices are the PEMs.This brief article aim to provide which next step to do and which one not to do when having a chassis alarm of fan failed on a PEM unit.
Alarm present:
show chassis alarms
2 alarms currently active
Alarm time Class Description
2024-01-24 19:33:23 UTC Minor PEM 0 Fan Failed
OR
2024-01-24 19:33:23 UTC Minor PEM 0 Fan Failure
Following logs are found in show log messages
Jan 24 06:58:22.003 2024 xxxxxx alarmd[13486]: %DAEMON-4: Alarm set: Pwr supply id=33554533, color=YELLOW, class=CHASSIS, reason=PEM 0 Fan Failed
Jan 24 06:58:22.003 2024 xxxxxx craftd[8546]: %DAEMON-4: Minor alarm set, PEM 0 Fan Failed
Jan 24 06:58:22.115 2024 xxxxxx chassisd[8541]: %DAEMON-5-CHASSISD_SNMP_TRAP6: SNMP trap generated: Power Supply failed (jnxContentsContainerIndex 2, jnxContentsL1Index 1, jnxContentsL2Index 0, jnxContentsL3Index 0, jnxContentsDescr PEM 0, jnxOperatingState 6)
You will notice all the PEMs are online:
PEM 0:
State: Online
DC input: Check (2 feed expected, 2 feed connected)
DC input: 48.0 V input (57500 mV)
Capacity: 4100 W (maximum 4100 W)
DC output: 513 W (zone 0, 9 A at 57 V, 12% of capacity)
PEM 1:
DC input: OK (2 feed expected, 2 feed connected)
DC input: 48.0 V input (57000 mV)
DC output: 570 W (zone 1, 10 A at 57 V, 13% of capacity)
PEM 2:
DC output: 399 W (zone 0, 7 A at 57 V, 9% of capacity)
PEM 3:
DC output: 627 W (zone 1, 11 A at 57 V, 15% of capacity)
System:
Zone 0:
Allocated power: 1472 W (2628 W remaining)
Actual usage: 912 W
Zone 1:
Allocated power: 1871 W (2229 W remaining)
Actual usage: 1197 W
Total system capacity: 8200 W (maximum 8200 W)
Total remaining power: 4857 W
This is a expected behavior as the power provided by the PEM is still working and the only damage component is the FAN of the PEM
We can try to reboot the router during a maintenance window, if the alarm is not cleared. PEM should be replaced.
This type of alarm is on most cases a direct hardware failure.
Bellow are actions to do:
1) Access PRsearch website, select your platform and OS used to discard alarm has been raised due to software bug.
2) Go to Juniper Portal to open a Service Request, collect below log captures, and ask for RMA for the unit specified on the logs.
>show log messages | match PEM
>show log chassisd | match PEM
RSI file: request support information | no-more
2024-01-25 : Article Created
2024-10-17 : dconejomora - Changing KB Visible To value for "Internal, Partner and Customer"
2025-01-03 : dconejomora - removing the below step from the KB, we should not instruct the customer to physically re-seat any FRU.
"The recommendation is that if the PEM has failed for the first time, we need to physically reset and monitor."
2025-04-15-Removed public only
2025-02-28: Added QFX and another variant of this log.