In the newer 19.4R3-S3.3 code when QFX10002 devices have more than 2 PEM’s slotted in the chassis and when the extra PEMs are not powered up or have no power feeds, a major chassis alarm is generated along with syslog and snmp traps. This behaviour is not seen on older release 17.3R2-S1.

Power Supply Traps are generated every hour when PEMs are present and input feeds are not connected. This creates high number of false alerts in customer's monitoring tool making it harder to be able to identify real problems in the network.

Below is the alarm shown in the RSI.

lab@qfx> show chassis alarms

2 alarms currently active

Alarm time Class Description

2023-08-18 04:52:29 UTC Major PEM 3 Not Powered <<<< 

2023-08-18 04:52:29 UTC Major PEM 1 Not Powered <<<<

syslog message:

Aug 25 07:27:35 qfx alarmd[8579]: Alarm set: PS SFXPC id=405798987, color=RED, class=CHASSIS, reason=PEM 3 Not Powered

Aug 25 07:30:47 qfx chassisd[7636]: CHASSISD_SNMP_TRAP6: SNMP trap generated: Power Supply failed (jnxContentsContainerIndex 2, jnxContentsL1Index 4, jnxContentsL2Index 0, jnxContentsL3Index 0, jnxContentsDescr Power Supply 3, jnxOperatingState 6)

 

From the below output, we can see that there are only two power supplies detected i.e PSU0/PSU2 in <show chassis hardware> output and PSU1/PSU3 are showing as absent (No power feeds or input power connected). 

lab@qfx> show chassis hardware detail no-forwarding

Hardware inventory:

Item    Version Part number Serial number  Description

Chassis        DG935    QFX10002-72Q

Pseudo CB 0

FPC 0   REV 35 750-XXXXXX AXXXXXXX   QFX10002-72Q

Power Supply 0 REV 04 740-xxxxx AXXXXXXXXXA  DC AFO 1600W PSU << Detected

Power Supply 2 REV 04 740-xxxxx BXXXXXXXXXB  DC AFO 1600W PSU << Detected

lab@qfx> show chassis environment no-forwarding

Class Item       Status  Measurement

Power FPC 0 Power Supply 0   OK   20 degrees C / 68 degrees F << Detected and powered on

      FPC 0 Power Supply 1   Absent <<<<< Module present in the slot however it is not powered

      FPC 0 Power Supply 2   OK   21 degrees C / 69 degrees F << Detected and powered on

      FPC 0 Power Supply 3   Absent <<<<< Module present in the slot however it is not powered
 

Power Supply Traps are generated every hour when PEMs are present and input feeds are not connected. This is day 1 behavior and same across other platforms as well. This is as per design. Major Alarm will be raised when the number of active PEMs/PSUs in the system is less than the configured "pem minimum" for the chassis. Default minimum number of PSUs for chassis is set to the max number of PSUs that the chassis supports. We can modify this by using the configuration statement “set chassis pem minimum <Minimum number of PEMs>”. It should avoid this alarm and SNMP Trap.

In addition to the the above configuration statement, you can configure the log messages to suppress power supply log messages 

set system syslog host <ip_address> match "!(.*Power Supply 1.*|.*Power Supply 3.*)"
set system syslog file messages match "!(.*STL library initialized.*|.*Power Supply 1.*|.*Power Supply 3.*)"
 

If in-use and not in-use PEM slots change (RMA for example), the configuration lines mentioned above need to be updated so the correct PEM syslogs are suppressed.

2023-08-28 - Initial publication
​​​​​​​2024-07-29 - Visibility modified