05-13-2015 12:07 AM - edited 03-07-2019 11:59 PM
im having a strange issue with a fiber connection between 4500-x and 6500 , it is a 10 gig connection with both LR sfb on both 4500 and 6500 and a sm cable , the issue is that the port on the 4500-x keep switching between connected and notconnected but the the port on the 6500 stays connected , when i check both interfaces on 4500 and 6500 it have gradually increasing crc errors and input errors on both switches , i tried replacing the cable , the sfb and the port with no avail , i tried switchport non-negotiate also tries setting it to trunk and access and still the same story , even when i connect the 4500-x with the same sfb to 2960-x sfb 10 gig as well there is no issues , also connecting the 2960-x to the 6500 shows no issues using the same 10 gig cable and sfb , please help as i ran out of solutions
05-13-2015 12:20 AM
have you tried to hardcode the speed and duplex settings , does it happen on all 10gig connections between the 65 and 45 or just the same interface each time
crc and inputs together usually infer there's a layer 1 issue , how far away are these devices from each other attenuation can be a cause for this as well , whats the reliability show on the interface when connected is it 255/255
you have ruled out sfp and cable by using different equipment so i would check these next
If there sfps that support DOM you can you can do a show interface transceiver
05-13-2015 12:32 AM
regarding the duplex i think u cant change it for sfb it is always full-duplex and yes i entered the bandwidth under the interface and it didnt work i even lowered the bandwidth , also yes i tried three ports on both switches and the same issue presist ,regarding cable im using a 1-1.5 meter back to back cable between the two devices , but i havnt checked the reliability.
05-13-2015 01:16 AM
yes sorry fibre you dont have duplex ,can you post the show interface for each side
the exact type of sfps and the ios currently in use in on 45
where are the crcs mostly incrementing on 45 or 65 side
the fact the issue does not exist using same cabling and sfps with 2960 & 6500 would indicate issue is on the 4500 whether its hardware or software support problem
05-19-2015 02:59 AM
the interface and the version of the 6500 is:
1-version >
05-19-2015 02:59 AM
They need to be LC-LC connections or
The ports can be configured with either SFP transceivers for 1-GB operation or SFP+ transceivers for 10-GB operation. Both transceivers use LC-type connectors (optical) or RJ-45 (copper).
05-19-2015 07:49 AM
Our connection is an LC (4500-x) to SC (4900). We didn't have an issue for the 1st 8 months in production. Is that due to software\hardware limitations, or is that a documented requirement?
05-19-2015 07:57 AM
That was taken out of a Cisco document for 4500 series switches 10GB requirements, i also came across it in another forum that they must be LC either side to achieve 10GB and SFP+ , you could check it with TAC but it looks to be hardware specific
05-19-2015 09:39 AM
markmalone2008,
I don't know where you got this information from, but we have several areas where we connect switches together using 10G, where one end is LC and the other end is SC. We have done it with multimode fiber and SR optics, as well as with singlemode fiber and LH optics. It's never been a problem.
Can you post a link to the Cisco document that says this isn't supported?
To the OP, can you post your port configurations? Also, when you say that the port(s) go down, how long do they remain down? Do they bounce and come back up?
Are you setting your ports to negotiate everything, or are you hard setting them to encapsulation (if it's an option), mode (access or trunk), turn off negotiation?
Have you tried configuring them as Layer 3 ports? If so, does the problem still occur?
-rb
05-19-2015 09:44 AM
baselzind,
Have you verified that your patch cable is certified for 10G operation?
Also, since you are using Single Mode optics, you need to be careful with optical power levels. With a short fiber optic cable, you may be overloading the optical receiver. In the long run this will burn out your SFP. In the short run, it may very well be causing the errors you are seeing.
It is very important to calculate an optical power budget for single mode connections, and to use attenuators to make sure that the receive levels are within specifications for the transceivers. Typically, attenuating it so the receive value is slightly above the middle range for the transceiver will be best.
-rb
06-03-2015 08:46 AM
Ron,
How do you modify the optical levels?
Fyi - this case has been escalated within Cisco and hope to have a game plan in the next day or two
05-19-2015 10:36 AM
I didnt say it was not supported and not the same for different switches just what i found online as the same cabling and sfps work with other switches fine and to check with TAC, this is one of the docs if your running sup7 series in 4500 series you need to use lc-lc
https://supportforums.cisco.com/discussion/11629431/catalyst-4510r-e-10gb-sfp-not-working
http://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst4500/hardware/configuration/notes/OL_23144.html#wp153042
05-19-2015 02:00 PM
markmalone2008,
Neither one of those links says that you need to use lc-lc. What they say is those particular switches have SFP optics, which by DEFAULT use LC connectors. Therefore, those switches can ONLY have LC connectors for those ports.
Conversely, SC connectors are found on X2 and XENPAK transceivers for 10G optics. So, Catalyst 6500 switch modules such as WS-X6708-10GE (X2) or WS-X6704-10GE (XENPAK) would have SC connectors for 10G. But as I said earlier, we have several connections using SC-LC, generally connecting an X2 module to a Service Provider 10G connection. In our cases, the cables are 15-20 meters long, and we still utilize attenuators to pad down the optical level on the receive side of each connection.
-rb
05-19-2015 12:42 PM
We are setting them as trunks, negotiation was not turned off. Although, I did just turn it off as it hasn't been tried yet. We have not attempted to set it us as a Layer 3 port.
The port does down, bounces for a few minutes and typically comes back up after 12-18 minutes. It then stays up for 5 + plus and occurs again. Now, this was fine for 8 months and then it started happening every 5 hours. We even had the 4500-x switches replaced.
Currently, there is nothing on the 4500-x except a trunk link to the 4900 and its still happening.
4500-x trunk config
interface TenGigabitEthernet1/1
switchport trunk native vlan 999
switchport trunk allowed vlan 99
switchport mode trunk
switchport nonegotiate
mtu 9198
4900 trunk config
interface TenGigabitEthernet3/8
switchport trunk native vlan 999
switchport trunk allowed vlan 99
switchport mode trunk
switchport nonegotiate
mtu 9198
05-19-2015 01:19 PM
ttcroziercisco,
How long are your fiber optic cables? If they are short, what is the requirement to use singlemode rather than multimode?
I highly suggest calculating an optical power budget for your connection, and pad the levels down as necessary. Also, combined with padding down the optical levels, test the connection with a different set of SFPs on both sides. If the problem is related to high optical levels, you may have damaged the SFPs.
-rb
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide