[f-nsp] Incoming CRC-Errors on remot side but no outgoing errors on my switch but there are WriteDrops.

[Gunther Stammwitz]

Hello Nico,

Thanks for your reply. Looks like you're up all night too :-) 
Any idea what might be the cause of the CRC errors?
We have already replace the fiber optic cables (yes: all of them) and
changed the switchports both on our local end and on the remote end.

On my second system that's built exactly the same way but has only half of
the traffic on it the drops are really low like 10 - 100 / hour but on this
one they're really really high. 

I think that my system is causing the problems since a second remote switch
is showing crc-errors.. Okay: only 5 errors in the last hours but this
should be zero, right.

What to do next?

Replace the module... But which one?


Gunther


> -----Ursprüngliche Nachricht-----
> Von: Nicolaj Kamensek [mailto:nk at accelerated.de] 
> Gesendet: Freitag, 16. Juni 2006 03:30
> An: Gunther Stammwitz
> Cc: foundry-nsp at puck.nether.net; 'Torsten Dobert'
> Betreff: Re: [f-nsp] Incoming CRC-Errors on remot side but no 
> outgoing errors on my switch but there are WriteDrops.
> 
> Gunther Stammwitz wrote:
> > Hello my dear colleagues,
> 
> Hi,
> 
> > I'm seeing some WriteDrops on the primary management module of my 
> > foundry BigIron 4000.
> 
> SSH at router>sh back
> ______________________________________________________________
> _________
> Slot  Mod   FreeQ    DMADrop    BPDrop    WriteDrop     Last
> --------------------------------------------------------------
> ---------
>   1   B0GMR4  3945       0          0      18409     D:99 H:22M:38S:3
>   2   B24E     900       0          0       7490     D:99 H:22M:38S:3
>   3   B8G      916       0          0          0          NEVER
> 
> And we're also having those drops on a different machine with 
> only a B8GMR3-A. In fact I've never seen a foundry with M3 or 
> M4(no experience with M1 and M2) without those drops.
> But unlike you we're not experiencing any CRC errors at all. 
> However, the WriteDrop counter might not indicate a problem, 
> the CRC errors in the opposite may.
> 
> Regards,
> Nico
> 
>