Description

This article provides guidance on addressing the "PEM [X] Load Sharing Failed" issue encountered on Juniper PTX/MX routers.

Symptoms

You will observe the below alarm on the device 

 

root@RE0> show chassis alarms

3 alarms currently active

Alarm time Class Description

2024-09-04 20:27:45 PDT Minor PEM 4 Load Sharing Failed

2024-09-04 20:27:30 PDT Minor PEM 1 Load Sharing Failed

2024-09-04 20:26:40 PDT Minor PEM 0 Load Sharing Failed

 

The router uses Power Entry Modules (PEMs) to manage power distribution across the system. The "Load Sharing Failed" message suggests that PEM [X] is unable to properly share the load, which can impact the router’s operation and reliability.

 

Common Causes due to which you can observe the alarm.

  1. PEM Hardware Failure: A malfunctioning PEM can lead to load sharing issues.
  2. Configuration Issues: Incorrect configuration settings can prevent effective load sharing between PEMs.
  3. Overload Conditions: The system may be drawing more power than the PEM can handle.
  4. Environmental Factors: Issues such as overheating can affect the performance of the PEM.

Solution

You can see 2 use cases for this issue, which is explained below in detail. 

Case 1:
If all the PEMs are nicely sharing the load, however the alarm remains, please follow the below procedure to resolve the issue. 

- First, you verify the status of the PEMs using the following command:

> show chassis power detail | no-more  

This command will show you that the power is distributing across each PEM evenly or not, if its distributing evenly across each PEM and you still observe the issue, please follow the below workaround to clear the alarm. 

- In this case, the issue is not impacting functionality. It is just a cosmetic appearance of alarm in system even though PEMs are functioning as per expectation.

- Workaround as below. 
1. Physically remove input and output source of PEMs and re-insert. If this doesn't clear the alarm, follow the 2nd step.
2. Physically reseat each PEM one-by-one. 
The above 2 steps will remove the load sharing failure alarm from the system.


Case 2:
If all the PEMs are not evenly distributing the Power among themselves, for example one PEM takes up 80% of load whereas the other PEM takes up 20%. In such case, the above procedure will not be going to help. 

- First, we need to identify which PEM module is taking up the maximum load, from the below command output. 

> show chassis power detail | no-more

root@RE0> show chassis power detail | no-more
Sep 18 21:19:12

PEM 0:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 624 W (zone 0, 52.00 A at 12.00 V, 23% of capacity)

PEM 1:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 1716 W (zone 0, 143.00 A at 12.00 V, 63% of capacity)

PEM 2:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 0 W (zone 0, 0.00 A at 12.00 V, 0% of capacity)

PEM 3:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 444 W (zone 0, 37.00 A at 12.00 V, 16% of capacity)

PEM 4:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 792 W (zone 0, 66.00 A at 12.00 V, 29% of capacity)

PEM 5:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 216 W (zone 0, 18.00 A at 12.00 V, 8% of capacity)

- Once you identify the PEM Module taking up the maximum load, you need to offline/remove the PEM from the device with the help of the site technician. 
- Once you remove the PEM with maximum load, the other PEM starts sharing the load equally among themselves. Post this observation you need to replace the PEM. 

root@RE0> show chassis power detail 

PEM 0:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 768 W (zone 0, 64.00 A at 12.00 V, 28% of capacity)

PEM 1:
  State:     Empty
  Input:     Absent

PEM 2:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 744 W (zone 0, 62.00 A at 12.00 V, 27% of capacity)

PEM 3:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 768 W (zone 0, 64.00 A at 12.00 V, 28% of capacity)

PEM 4:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 756 W (zone 0, 63.00 A at 12.00 V, 28% of capacity)

PEM 5:
  State:     Online
  Capacity:  2700 W (maximum 2700 W)
  AC input:  OK (Both feed expected, Both feed connected)
  DC output: 768 W (zone 0, 64.00 A at 12.00 V, 28% of capacity)

- If the above doesn't work, you need identify the PEM with minimum load and offline/remove that PEM to check the status of the load sharing. 
- If in this case, after identifying and removing the faulty PEM you still observe the "Load Sharing Failed" alarm on the PEM you need to follow the workaround mentioned in Case 1, that will clear the alarm. 

If the issue persists after performing the above steps, consider reaching out to Juniper support for further assistance.

Modification History

2024-09-19 : Article Created