The port flapping issue is definitely real and I am researching this for a long time. I have tried switching ports, cables, it does not matter, in 99% of cases its not cables, I am tired by how many people mention cables. In my life, the cable issues have hints that look totally different, for example, a bad cable will NEVER produce such an effect like being able to work for 1 month without a single flap and then suddenly start flapping non-stop, independently of traffic, where re-plugging does nothing (or only for a short time), the only thing that helps (for much longer, but still limited time) is power cycling the RB5009. What is more interesting, when flapping starts, even the reboot does not help for long, only power cycling manages to restore it for longer.
Also as an electronics engineer I can assure that at short lengths the cable quality does not even matter, it can be absolute junk and still work perfectly because it manages to not only meet, but even exceed the necessary signal quality/strength tolerances. It is almost impossible for a bad cable to create ANY problems if its short enough.
Anyway, in my case its the 1Gbit connection between RB5009 and Netgear GC110P managed switch. Sometimes it can work for a month, two or even more, but when the flapping starts the only solution is power cycling RB5009. Nothing else really helps, neither power cycling the Netgear switch, nor re-plugging the connectors on both ends, at least not for long. I have even tried setting the RB5009 port to 100Mbit when I noticed that flapping started, without power cycling, it worked for some time, but then started flapping even on 100Mbit, so it definitely is not the solution. Only power cycling helps.
Another curious thing that I noticed on 100Mbit, the link LED on Netgear switch can go dark after some time, but the link will still be working and not flapping. Its enough to re-plug the UTP cable in order to wake up the LED. It looks like some part of the ethernet auto-negotiation process in RB5009 stops working or starts working incorrectly. The signal quality reasons are definitely not the case here, especially when talking about 100Mbit where the cabling requirements for everything to work perfectly are simply laughable and easily _overachieved_ even with cabling that is beyond poor.
Another possible reason that I'm thinking of is maybe it is some timing/clocking/pll issue with the switch chip 88E6393X that is used in RB5009, like maybe the clocking gets skewed over time, becomes more sensitive to something, that would explain why does it manage to work perfectly for quite a long time until something finally provokes it to become unstable, from where only the power cycling really helps, which would also mean that a soft reboot does not reset the chip internals where the clocking error accumulates in some way, and only reinitializes it. Hopefully correctable by software and not a chip/PCB design flaw.
Also as an electronics engineer I can assure that at short lengths the cable quality does not even matter, it can be absolute junk and still work perfectly because it manages to not only meet, but even exceed the necessary signal quality/strength tolerances. It is almost impossible for a bad cable to create ANY problems if its short enough.
Anyway, in my case its the 1Gbit connection between RB5009 and Netgear GC110P managed switch. Sometimes it can work for a month, two or even more, but when the flapping starts the only solution is power cycling RB5009. Nothing else really helps, neither power cycling the Netgear switch, nor re-plugging the connectors on both ends, at least not for long. I have even tried setting the RB5009 port to 100Mbit when I noticed that flapping started, without power cycling, it worked for some time, but then started flapping even on 100Mbit, so it definitely is not the solution. Only power cycling helps.
Another curious thing that I noticed on 100Mbit, the link LED on Netgear switch can go dark after some time, but the link will still be working and not flapping. Its enough to re-plug the UTP cable in order to wake up the LED. It looks like some part of the ethernet auto-negotiation process in RB5009 stops working or starts working incorrectly. The signal quality reasons are definitely not the case here, especially when talking about 100Mbit where the cabling requirements for everything to work perfectly are simply laughable and easily _overachieved_ even with cabling that is beyond poor.
Another possible reason that I'm thinking of is maybe it is some timing/clocking/pll issue with the switch chip 88E6393X that is used in RB5009, like maybe the clocking gets skewed over time, becomes more sensitive to something, that would explain why does it manage to work perfectly for quite a long time until something finally provokes it to become unstable, from where only the power cycling really helps, which would also mean that a soft reboot does not reset the chip internals where the clocking error accumulates in some way, and only reinitializes it. Hopefully correctable by software and not a chip/PCB design flaw.
Statistics: Posted by yozas — Fri May 02, 2025 1:47 am