Description

PFE process crash is observed on Junos EX4300 platforms

Symptoms

On Junos EX4300 platforms, a PFE process crash could be observed due to a rare timing issue that led to traffic loss during the time of the crash.

Below error log could be observed during the time of issue:

pfex: %USER-0: pfex_watchdog pfex watchdog triggered!!!!!!!!

Also, the "show system core-dumps" command can be used to check the core files.

 

Solution

In a rare scenario, due to timing issues, the Packet Forwarding Engine (PFE) crash is observed on Junos EX4300 platforms. This causes traffic loss until the PFE comes up.
This issue might be seen if the following conditions are met:
* On Junos EX4300 platforms
* Not applicable for EX4300-48mp


This is a timing issue: 
- Some high priority process make pfe ukern htread to starve
- When pfe ukern_thread never get to run cpu, there is a false positive in watchdog logic in pfe causing the pfe to panic and core.

This issue is addressed through PR1720219

Modification History

2024-09-23 : Article Created