ENC errors

[Thomas Kemp_1]

We have a fabric that consists of SW2800s, SW3900s, and SW12000s. I am seeing a large amount of ENC errors on most ports, across most switches. All ports are set to autonegotiate on the switches, including the ISLs. I have read in Brocade material that locking down ports to the speed of whatever is plugged into them can help inhibit ENC errors by combating autonegotiation problems. Can anyone confirm this?

[Jan-Erik Karlsson_2 · Accepted solution]

I can confirm that it is a recomendation to lock down the ports to the actual speed it should run in.
I have a SAN with two separat fabrics and I have all ports locked to 2GB except for thoose used for HSG80, they are set to 1GB, even thoose ports not used.

I don't see that problem you have.

Regards

[Matthias Rettl]

hy.

i integrated several brocade fabrics in the last months and every switch shows a huge amount of ENC OUT errors like in the log i attached!


i talked to a "guru" from brocade and he assured me, that you can ignore this errors, as long as you do not have a real problem. normally an ENC OUT error has something to do with physical problems (cable, sfp)... but in none of my configurations i do really have a problem. despite the ENC OUT errors.

maybe brocade will change this error logging in a later fabric os release.

regards,
matthias

[Thomas Kemp_1]

I spoke to two HP storage engineers, and they both highly suggested that the ports be locked down to inhibit ENC errors. Their major point for locking down the ports was to relieve the ASICs from the task of processing millions of bogus ENC errors caused by possible autonegotiation issues. By stopping the errors, you relieve the ASICs in switch from processing these errors, and therefore potentially increase throughput on the switch.

Thanks for both replies, and confirmations.

[Matthias Rettl]

hi!

try, but i don't think you'll get better results. there's also a recommendation to disable all unused ports on the switches... try this and watch the result (a blinking switch). just to give you a touch about recommendations of hp :)

[Thomas Kemp_1]

At a previous job site, we locked down the ports, but I could not remeber why. :) We didn't get high ENC errors unless a path was failing, which was expected.

My other concern is if you have a failing or degrading path, some of the indicators such as ENC errors would be ignored if most ports have high ENC errors. I guess I just don't like messy switches and things.

[Matthias Rettl]

yes, i remember to have the first ENC OUT error floods with OS 3.1.?... in earlier versions there where none.

my brocade-spezi told me, and this really differs to the hp guys, NOT to lock down the speed of the ports... so, whom could you believe? from my experiences i ignore the errors, i do not lockdown ports and i do not disable them... and i do not always do things, hp guys tell me *g*

good luck!
matthias