03-05-2020 03:26 PM
Does anyone have experience with massive CRC errors on one physical port of a LAG?
Is it possible that a C5 is not able to handle this kind of issue and loos the complete LAG Memberships?
03-22-2020 04:01 PM
Hello Brian,
We plan to replace MGBICs and enable ”single port lag” next downtime.
Lag.0.2 is down until then.
But without single port lag enabled, there is the following scenario.
Case 1) ge.2.23 is down and ge.2.24 is down lag.0.2 is down
Case 2) ge.2.23 is up and ge.2.24 is down lag.0.2 is down
Case 3) ge.2.23 is up and ge.2.24 is up lag.0.2 is up
Case 4) ge.2.23 is up and ge.2.24 is down lag.0.2 is up
Case 5) ge.2.23 is up and ge.2.24 is up lag.0.2 is up
Case 6) ge.2.23 is down and ge.2.24 is up lag.0.2 is up
Case 7) ge.2.23 is down and ge.2.24 is down lag.0.2 is down
Case ge.2.23 is down and ge.2.24 is up lag.0.2is down
Case 9) ge.2.23 is up and ge.2.24 is up lag.0.2 is up
There should be no problem when one port is no member of the LAG if the LAG has established before.
The C5 was connected with two LAGs.
We had massive problems after ge.2.24 on lag.0.2 had CRC errors.
Also problems with lag.0.1 packed forwarding.
It is working lag.0.2 links down.
We will see if all is Ok next time.
Thank you for answering.
Regards
Herbert
03-22-2020 03:04 PM
You can enable single port lag, so that if your lag gets down to one port up and running, then it will still operate as lag.0.x.
However you will need to remedy the crc errors otherwise performance will be spotty regardless.
03-22-2020 02:42 PM
Thank you Matthew for your answer.
I know this. Question is if massive CRC errors on one physical port and therefore missing LACP BDUs can bring the whole LAG to fail?
Can it be that there also other LAGs on the C5 are effected in this case?
Regards
Herbert
03-21-2020 06:36 AM
I have performed LAGs on C5’s to many other devices, including Cisco and other XOS devices without issues. If you are receiving CRC errors you may want to check L1, such as dirty fiber or a bad cable or bad gbic.