Regular CRC errors for 5-10 minutes for a couple days and then nothing for 4-5 days..

[johnjutte]

I have particular 10Gb link between an MS-125-48LP and MS-125-24P that is reporting brief CRC errors on loose but almost consistent basis. In the last month here's what I'm seeing:

11/2 - Switch installed no issues

11/8 - 2:46p to 2:56p - CRC errors side A - played wait and see. 

11/9 - 11:18a to 11:22a - CRC errors side A - cleared before noticed

11/16 - 3:59p to 4:06p - CRC errors side A - I took action. 

11/16 - 4:06p to 4:21p - replaced SFP+ modules and fiber cables on both sides

11/17 - 10:03a to 10:07a - CRC errors side A - cleared before I noticed

11/23 - 3:45p to 3:50p - CRC errors side A - vacation. 

11/24 - 11:56a to 12:00p - CRC errors side A - vacation.

11/29 - 2:30p to 2:35p - CRC errors side A - saw it this morning

11/30 - 12:05p to 12:10p -CRC errors side A - moved to different fiber pair with new module/cables

11/30 - 3:05p to 3:15p - CRC errors side B

While not precise.. there is some consistency with later in the day on Mon/Tues and then late morning the day after. The rest of week the link seems fine or at least not enough issues to alert. I understand CRC errors to mean a physical layer issue hence changing modules, cables, fiber strands, and ports. Are there other things I can check that might cause this? Is environment interference possible (outside RF or something)? Could support maybe dig deeper? Or should I try more hardware/get my fiber between closets checked? 

Devices on the switch uplinked through this connection never complain - includes VOIP devices and a MR36 with 10-15 devices connected all the time.  

[Brash]

You're bang on that CRC's are a Layer 1 or Layer 2 issue.
Most commonly it's dirty/loose/broken cables, or occasionally a bad SFP/switchport. It can also be due to MTU mismatch.
If it's a 10Gb fiber link, interference is highly unlikely as fiber is relatively unaffected by RF as opposed to copper.

In terms of the consistency in timing, a couple of things to note are:
 - Higher CRC counts will generally correlate with higher network traffic (more packets = more likely for one to be corrupted). Is it possible that at these times the link had a higher utilization in general?
 - Do you have any scheduled traffic around those times where higher MTU data is being sent (and where a mismatch might be present in the network)?

In any circumstance, I would definitely suggest starting with swapping out the cable and continue troubleshooting up the stack from there.

[johnjutte]

I had more, inconsistent, CRC errors overnight tonight. 

I changed modules, patch cables, and ports a few times now. I guess my fiber between closets is likely the issue (dirty or damaged). 

[johnjutte]

Well, so I might be stupid...
The link in question is OM1 and about 395ft long. A bit outside the range of 10Gb, so the link connected but signal is weak. I suspected this, and I was trying to some LRM modules with conditioning cables to see if I could be better signal. In a rush, not paying attention, I connect the conditioning cable backwards on 1 side (conditioner into patch instead of module).  

I flipped the conditioner around and so far so good... yea so, my bad guys. Thanks for input. 

[Brash]

Good find.

Certainly better than having to re-run cables.