Error Counters in Statistics (Network Port in Fabric Interconnect)

[andyblada1]

Hello UCS Champs,

Need your quick help on identify the cause of FCS & RCV errors. 

I am getting FCS & RCV errors on network port of fabric interconnect. Is this normal or there is some problem.  Will it lead to packet loss??

I didn’t find any concrete information on below errors in UCSM. Thought maybe you are aware of it.

Refer attached screenshot for more insight.

[Wes Austin]

There are only 47, so not likely causing any issues. However, if the number steadily increases, it could lead to problems down the road.

If this is an ethernet uplink, you need to go to the switch that is connected and trace the errors accordingly.

[andyblada1]

Hello Wes, 

Thnx for your quick reply!

Could you please put more light on the minimum and maximum count??

Till what number i should't be worry and worry number ?? just for idea so that i can start working on them proactively.

In other UCS infra, fcs & rcv error reached to 300 . Should I engage my network team to investigate it further.

Cheers 

Andy!!

[Wes Austin]

CRC/FCS errors can be caused by number of different things, but typically they are hardware and/or port configuration related. 

Seeing 10-20 errors here and there is not typically cause for concern. If you start seeings numbers in the upper hundreds/thousands, it definitely warrants further investigation.

From the NXOS shell on your FI, if you run 'show interface counters errors' you can see what network uplinks are transmitting the errors and use the 'show cdp neighbors command' to see what interfaces on the upstream switch require investigate.

[andyblada1]

Thanks Mate !! Let me give a try.
Apart from this, do you have any link to refer, if i want to study more on these errors reflecting in statistics in UCSM.

Cheers
Andy !!

[andyblada1]

Hi Wes,

I tried running command "show interface counters errors" on both the FI's.

Network Ports in FI-B showing FCS and RCV erros more than 300 & 400 respectively. 

Server Ports in FI-A showing FCS and RCV erros more than 800 & 1000 respectively.

Refer attached document for more details. 

Does it needs attention??

Also, i tried command "show cdp neighbors" but it is showing details of clustered FI, not Northbound switch. Seems it is disabled at upper layer.

Cheers

Andy !! 

[Wes Austin]

If the errors are not incrementing, they likely do not require attention and could be historical. You can use the "clear counters" command from the nxos shell to clear those and monitor accordingly.

If you are receiving CRC/FCS errors from the upstream network, you likely have a physical cable/SFP problem northbound that you need to address OR you have a MTU mismatch.

If you are receiving these errors from the server links, then you could have a bad adapter OR MTU violation on the hostside, sending jumbo frames up to the network erroneously.

I would open a TAC case for investigation and if you do, please send me the SR so I can take a look.