For more details, please see ourCookie Policy.


Fibre Channel (SAN)

Reply
Occasional Contributor
Posts: 6
Registered: ‎11-10-2010

New counter field pcs_err

We have 8 backend paths from each of a set of switches to a flash array with Load Balancing on the Op/Sys.

 

3 of  those backend paths on each switch report pcs_err counts around 1000. All other counters are 0 or very low.

 

I've searched and found nothing about this new counter field except that it is 16GB + only and is for the Physical Coding Sublayer.

 

So, I am wondering if these counts indicate a physical SAN problem or possibly a storage, switch or HBA configuration problem ?

 

 

Valued Contributor
Posts: 559
Registered: ‎03-20-2011

Re: New counter field pcs_err

I'm not sure what's the meaning of this counter. You could ask your Brocade or disk array support about it. But from experience, the problem indicator is the regular or at least periodic growth of the counter. If it doesn't grow - everything is OK.
External Moderator
Posts: 5,541
Registered: ‎02-23-2004

Re: New counter field pcs_err

as per my experience, the pcs_err refert to ITW = Invalid Transmission Word and can occur due a Incoding_in and Incoding_out Violations. Encoding violation occur only at 8G or lower, PCS Violation at 10G or higher speed.

TechHelp24
Valued Contributor
Posts: 559
Registered: ‎03-20-2011

Re: New counter field pcs_err

That's interesting, thank you for sharing!

Having this in mind, my recommendation stays the same:
- If the counter is only growing during the initial phase of the link setup, but then doesn't grow during the normal production data transfer, I'd say this is some transitory event and it shouldn't have a lot of concerns.
- If the counter keeps growing further, the issue should be investigated, especially because of this last update from Antonio which links this particular counter to the ITW and/or encoding like conditions - that's serious.
New Contributor
Posts: 4
Registered: ‎07-05-2005

Re: New counter field pcs_err

After replacing the sfp (degraded to 8gb, lots of enc_out errors), we are now seeing lots of pcs_err on the port. It is 1 of 2 trunked ISLs and the other reports zero issues. The sfp came back at 16GB and the trunk is now OK at 32GB. Neither port shows any errors but the pcs_errs. Google has been no help, and "you probably have a problem with the line" means nothing to me as far as determining whether I'm having a real issue that impacts anything.

 

These are the first instances of 16GB that we are implementing, and I would appreciate any help regarding this new error field. Thanks

 

This is my trunk (portstatsclear run on p28 more than once):

          frames       enc    crc    crc    too    too    bad    enc   disc   link   loss   loss   frjt   fbsy  c3timeout    pcs
             tx     rx      in    err    g_eof  shrt   long   eof    out   c3      fail    sync   sig                    tx    rx          err

 28:   13.5m 514      0      0      0      0      0         0      0        0      0        0        0      0      0      0      0          2.3m
 29:    2.3g   2.0g     0      0      0      0      0         0      0        0      0        0        0      0      0      0      0           0  

Frequent Contributor
Posts: 134
Registered: ‎03-20-2018

Re: New counter field pcs_err

Run sfpshow 28. Look at the RX and TX Power. They should be roughly similar. If RX Power is much lower than TX Power, it could indicate a line issue.

New Contributor
Posts: 4
Registered: ‎07-05-2005

Re: New counter field pcs_err

And I think we have found the problem... when I checked the other end of the trunk, I noticed the stats for the 'bad' port were basically one-way, and it was the reverse of the other switch. Seems like a physical connection issue, or an sfp in need of a cleaning. 

 

From switch with pcs_errs:

28: 14.0m  528 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2.4m
29: 2.3g    2.0g 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

Other side:

20: 1     21.1k 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
21: 2.3k 2.3k  0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Contributor
Posts: 50
Registered: ‎05-26-2010

Re: New counter field pcs_err

[ Edited ]

The pcs_err counter is specific to 10Gb and 16Gb fibre channel connectivity.  PCS = Physical Coding Sublayer (PCS) block errors. It increments when encoding violations on 10 Gbps or 16 Gbps ports occur.  In all cases I've seen it is caused by physical layer (SFP or cabling including any patch panels in the cable run.) 

The comments and advise included are the author's and do not mirror the views and opinions of any company. Any comments, advise, statements or references are without warranty, liability or claims of expertise. You should seek the assistance of the professional entity for subject matter being pursued.

Join the Broadcom Community

Get quick and easy access to valuable resources across the Broadcom Community Network.