Nexus 7700 and single-bit ECC error

[roysm]

Hi

I have noticed the following error being logged on a Nexus 7700 switch:

%MODULE-4-MOD_WARNING: Module 5 (Serial number: JAE20470361) reported warning Ethernet5/25-28 due to single-bit ECC error in device DEV_CLP_L3 (device error 0xcaf06600)

The error appears to logging every 1-2 seconds. Can anyone tell me what this might indicate? Am I looking at a hardware fault on the port or module? 

This is a  Nexus 7700 6-port switch running n7700-s2-dk9.8.4.2.bin. 

The module in question is a N77-F248XP-23E card. Ports Ethernet5/25 and 26 each have 10G modules while ports 27 & 28 do not have anything connected and no modules installed. Also, ports 5/25 and 5/26 are in different Etherchannels along with the corresponding ports 6/25 and 6/26. However, I do not see any errors for module 6.

Considering, that only 2 of the reported ports are actually being used and the other ports on port 6 are fine, I'm looking for some information on what this error might refer to before replacing anything but my searching has not found anything.

Has anyone seen this error before? Any help would be appreciated?

Many Thanks

Roy

[f00z]

Usually the module has to be reloaded, and if they continue again it's a hardware fault.  Single bit ECC errors happen randomly sometimes but shouldn't happen a lot . If they happen more than a few times per year something could be wrong with the hardware. I believe the DEV_CLP_L3 Is the Clipper ASIC L3 lookup memory.    It happens to every device at some point and some log it and some don't (unless it's excessive).  Like the 6500/6800's don't log it unless service internal is running (i.e. it's the same thing as parity error)

There's also a bug report on these ( as they shouldn't be syslogging every time but only if they are incrementing quickly )

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCve70443/

I would reload the module and see if it continues.

[roysm]

The bug report is interesting, although I am running nxos v.8.4(2), which is later than the fixed versions referred to. 

I have reloaded the module with no change. Due to vast amount of errors being logged, I have to lean towards this being a hardware fault, so I have raised a TAC call. 

Thanks
Roy