This topic has been marked solved and closed to new posts due to inactivity. We hope you'll join the conversation by posting to an open topic or starting a new one.
In Switch1, ports 50 & 52 (bottom SFP ports) are setup in a LAG.
In Switch2, ports 25 & 26 are setup in a LAG.
When I connected the fiber cables (nothing else connected to Switch2), I experience weird network issues:
Some devices on Switch1 could only communicate on the LAN and couldn't hit the FW and get out to the Internet.
For devices on Switch1 that could still get to the Internet, running a continuous ping would result in frequent dropped packets.
The ports on Switch2 would eventually disable.
I disconnect one of the two fiber cables, the issues resolve.
FWIW, I have this same configuration on different floors of our building. The 26 port switches are new to support a new camera system, but in each case are connected to a different 52 port switch, but in the same fashion. They both show similar behaviors when both fiber cables are connected.
As I'm looking at the cable and module specs, did I get the wrong cables? Oddly, at a different location, I have two GS752TPv2 switches connected with the same modules and style cables (just twice as long, ~3ft).
In this configuration, should I have gotten these instead:
This are MMF (Multi Mode) modules, while the fiber is SMF (single mode). A SMF fiber core is just 9 um (in the 125 um total glass cover), while MMF is 62.5um core (OM1) resp. 50um core (OM2 or better) both in 125 um glass cover. So yes, wrong patch cable.
This are MMF (Multi Mode) modules, while the fiber is SMF (single mode). A SMF fiber core is just 9 um (in the 125 um total glass cover), while MMF is 62.5um core (OM1) resp. 50um core (OM2 or better) both in 125 um glass cover. So yes, wrong patch cable.
However, any thoughts on why this seemingly works on two connected GS752TPv2 switches?
Same modules, same cables (just longer, ~1m/3ft). That's why I purchased the same items.
Or is this a case where maybe it's not working like I'd think, and I'm not getting the benefits of LAG but the switch has error correcting to where it's not disrupting normal communications? I don't know, I was just surprised that it appears to be working in two other instances. I've ordered the multimode cables.
However, any thoughts on why this seemingly works on two connected GS752TPv2 switches?
Same modules, same cables (just longer, ~1m/3ft). That's why I purchased the same items.
There are some theories and practical hints about mixing MMF modules and MMF and SMF fibers around. The "trick" is to use longer (I remember having used some 10 meters of SMF) to "focus" the light while reducing reflections from the "outer" mechanical glass carrier. Think bout that both types re built on a 125 um outer carrier glass "tube" holding the effective 9 um (SMF) resp 50/62.5 um inner thin fiber effectively used for the transmission/reception.
The fiber we see is always the complete 125 um diameter glass "tube", while the effective transmission and TX/RX optics is optimized for 9 um resp. 50/62.5 um.
This is just not cooperating, I'm sure it's simple, like the cable lengths.
Here is the current setup, switch 4&5 are *new* to support a security system.
I realize this config is far from optimal (e.g., Switch2 dies we're toast; I want everything in the basement, only a single RJ45 run downstairs, etc.) but just haven't had the time or resources to break up all these failure points, but I digress. Until introducing the new switches, everything had been going along swimmingly. Even from a VLAN, QoS and optimizations, considering my core competency isn't network engineering, we have plenty of bandwidth, Teams video calling is crystal clear, and razor sharp, VoIP calls as well...Basically, trying to illustrate the network is healthy.
Add the new switches with the SFP modules (NETGEAR AGM731F Compatible SFP 1000BASE-SX 850nm 550m DOM Duplex LC/UPC MMF Optical Transceiver) and patch cables (0.5m (1.6ft) LC UPC to LC UPC Duplex OM4 Multimode Riser (OFNR) 2.0mm Fiber Optic), and I just start getting network issues, primarily with devices hanging off Switch1 and then Switch5 will disable the SFP modules. Nothing as of yet is connected to the new switches aside from the LAG. I disconnect one of the two cables from Switch4 and issues with Switch1 clears. You mentioned mixing cables, did you mean literally within the same infrastructure (so in my case, all 5 switches need to be connected with the same cable type)? I realize mixing cables between the same switches would be unwise. Would I really be doing myself a favor if I got rid of the single mode fiber between the GS752TPv2 switches? Or is this just, "Hey dummy, get a 1m cable for the new switches?" Would you recommend orange which is what OM1/OM2?
I overlooked enabling spanning tree on the *new* switches, so the LAG was creating a loop. However, when I enabled STP on the new switches, the problem started occurring again. The new switches currently have just a single link on one of the 2 SFP ports. I certainly understand why I had the issue with 2 links and no STP, a broadcast storm/loop with no way to mitigate it. But is the opposite also true if it's enabled and there's not at least 2 physical links between switches? Basically, I was going to enable it, go to the office tomorrow and connect the second link for the LAG. but as soon as I did that, I start getting connectivity issues again.
I've also been reading one of the devices being the root through election. Is switch2 likely to become that based on where it is in the topology? It's also the only switch connected directly to the FW/router.
Before the LAG wiil become active, these two conections are individual network links - in case something misconfigured, this could easily lead to a loop situation.
I'm going to go from "left to right," so to speak relative to the diagram of our switches.
SW4, LAG1 (to SW2), Port State = Manual Forwarding, Cost = 10000
SW1, LAG1 (to SW2), Port State = Forwarding, Cost = 20000
SW1, LAG2 (to SW4), Port State = Forwarding, Cost = 10000
SW2, LAG1 (to SW1), Port State = Forwarding, Cost = 20000
SW2, LAG2 (to SW3), Port State = Forwarding, Cost = 10000
SW3, LAG1 (to SW2), Port State = Forwarding, Cost = 10000
SW3, LAG2 (to SW5), Port State = Forwarding, Cost = 10000
SW5, LAG1 (to SW3), Port State = Manual Forwarding, Cost = 10000
Based on what I'm reading STP uses the path with the least cost, so it seems that would be why I start seeing weird network behavior. I didn't enter these values, but is it possible the switch configured them at some point previously and can't update the values on its own? Any thoughts why it might be 20000? All ports are connected at 1GB Full Duplex, just a different medium (ethernet versus fiber).
Should I manually change the value of the respective LAG on SW1 and SW2 to 10000? That's how SW3 is configured, only difference is both LAGs are fiber.
Can you try enabling the storm control of the switch for broadcast and select trap for control action. You may check page 289-292 from the link below to be guided.
