10gig optic port goes down with remote fault, needs power down and up to get it back once a while

[Mary_Peng]

I have an extreme switch recently configured, the optical port is not working very good. The link state is always active for a while(couple of hours to couple of days), and then change to link state "ready" and never come back until I power the whole switch down and up.

Recent log, related to this issue is :
09/29/2015 22:05:50.48 Port 45 link down - remote fault 09/29/2015 22:05:50.47 Port 46 link down - remote fault
09/29/2015 07:02:32.41 Port 45 link UP at speed 10 Gbps and full-duplex
09/29/2015 07:02:32.40 Port 46 link UP at speed 10 Gbps and full-duplex
09/29/2015 07:02:27.37 Port 45 link down - Local fault
09/29/2015 07:02:27.36 Port 46 link down - Local fault

Seems like the local fault can be recovered by the switch itself, but if the remote fault occurs, the port won't come back. Does anybody know what the remote fault really means?

Switch Type: X670V-48t
EXOS version: 15.2.2.7

detailed transceiver information on the extreme switch side:
SFP or SFP+: SFP+
Signal: present
TX Fault: no
SFP/SFP+ Vendor: AVAGO
SFP/SFP+ Part Number: AFBR-709SMZ
SFP/SFP+ Serial Number: AD15163022E
SFP/SFP+ Manufacture Date: 150413
SFP/SFP+ Type: SFP
Connector: LC
Type: SR
Supported: no

Wavelength: 850
GBIC supports DDMI. MonitorType: 68

On the other end, we are connecting to EXFO QA-805 10 gig optic ports with transceivers supported by them.

Thanks!

[Mary_Peng]

Hi AlexandrP, I'm not very familiar with stack ports, is it set by default? I'm not sure if we are using stack port or not.

I tried the follow command for setting debounce timer:
X670V-48t.1 # configure port 45 debounce time 4000 ^
%% Invalid input detected at '^' marker.

But the command is not working for my console.

[EtherMAN]

need to get light readings if you can.. I would guess you are also using type 3 MM patches?? We have found if there is a bad patch or a single high loss in one fiber it will also randomly go down and you have to pull and reseat the sfp+ to get the link back up... We have also seen that some of the SFP+ just don't electrically seat properly and can be a bit flakey to keep stable

[Mary_Peng]

Thanks for your reply. I'm using MM50/125 (OM3) fiber patches and SFP+ as stated above. I think it might be the issue, if the following experiment is correct: I loop back from port 45 to port 46 using the fiber patch, the link is down after a while with "local fault" in event log and no way to up again without a cold reboot.

[Patrick_Voss]

Hi Mary,

You are correct. I don't see any errors. The errors do reset after a switch reboot and if the port never became active it wouldn't receive any errors. Can you try taking a known working GBIC and replace the one having issues. Check both sides as well, this could be related to the device on the other side.

[Mary_Peng]

Hi Patrick,

I heard other teams in our company use the same GBIC with another extreme switch with no problem. I didn't reboot the switch, and still kept it at the 'ready' state when doing the show command. Is there some other command that I can try to get errors happened on those ports?

The device on the other side don't show any error and I don't need to reboot it to get the port back. You are right, I will also contact with them to see if they know about this issue. But for any reason the port on extreme switch got an error, I think it should restart or active itself, instead of in state 'ready' forever.

Thanks,
Mary