PFE (Packet Forwarding Engine) board is reporting CCHIP abnormal discard errors.
The PFE:.*CCHIP: .*abnormal discard seen.* message is seen when the C-chip is discarding packets abnormally. This chip is usually found on the forwarding fabric component of the router.
The message will look similar to one of the following examples.
CCHIP: abnormal discards seen - check "show pfe statistics traffic"
CCHIP: prolonged abnormal discard seen - check "show pfe statistics traffic"
This message can be caused by faulty hardware.
There are also some rare software errors that can cause this message.
The error message exhorts you to check show pfe statistics traffic. This will list traffic discards; but it can still be difficult to say what is causing it. It is a good idea to run this command a few times and see which statistics are incrementing and by how much. The same goes for show pfe statistics error.
If you check the syslog messages (show log messages), you may notice a lot of the following kind of messages, along with the CCHIP message:
NH: Aggregate nh (370) with unknown nh in list
NH: invalid nh (370) for bulk statistics
This could indicate a software error. This error is not serious. There should not be any traffic impact; the errors are logged incorrectly due to the sequence of various statistical checks.
If the accompanying next-hop (nh) errors are not seen and/or you are experiencing a loss of traffic or services, then the issue is probably a hardware error. During a maintenance window (as it will disrupt traffic), try rebooting the router to re-initialize the memory. This should clear out the pfe statistics. Monitor this for incrementing discard statistics. If they continue to increment and you see the “abnormal discards” error again, then replace the component.
Perform the following steps:
- If you are seeing these errors along with a lot of “NH” messages, upgrade to Juniper recommended version or higher.
- During a maintenance window, as this step will disrupt traffic, reboot the router to clear out the memory.
- If the messages continue or appear again quickly, replace the component that is logging the errors.