Description

This article explains the unexpected behaviour of (but not limited to ) contineous fabric error condition when non-vmhost image is used to upgrade the NG-RE.

Symptoms

When customer tries to upgrade NG-RE with non-vmhost image the upgrade goes through without any issue if the existing vmhost version is compatiable with the junos upgrade version however this may show some unexpected behaviour like Fabric error and plane issues of FPC offline issue.

root@re0-lab121> show chassis alarms
2 alarms currently active
Alarm time               Class  Description
2024-03-19 09:26:41 UTC  Minor  Check CB 1 Fabric Chip 0
2024-03-19 09:23:06 UTC  Minor  Check CB 0 Fabric Chip 0

root@re0-lab121> show chassis fabric summary
Plane   State    Uptime
 0      Check    1 hour, 12 minutes, 44 seconds
 1      Check    1 hour, 12 minutes, 44 seconds
 2      Online   1 hour, 12 minutes, 9 seconds
 3      Check    1 hour, 12 minutes, 9 seconds
 4      Check    44 minutes, 20 seconds
 5      Online   44 minutes, 18 seconds

Solution

If the above errors are showing frequently and the alarm keeps coming back after replacing the suspected faulty component then this could be due to the the upgrade performed with wrong image which would cause a disparity between the VMhost image and Junos image.

Junos vmhost image is always bundled with the most compatible VMhost image and when used for upgrade on NG-RE it upgrade the vmhost simultaneously to ensure that the compatibility remain the same however sometime using the non vmhost image by mistake can cause issues.

In order to fix this issue upgrade the RE again with correct vmhost image and that should stop the errors.
 

Modification History

2024-05-13 : Article Created