Hmm I never had a problem getting them started. FWIW I'm getting 1-2 ms from my desktop, through my router/eptp/switches at the tower down to our BMU.
I also don't think 20ms is that big a deal for most customers. If they get say 60ms from their machine to most anything in a datacenter they should be better off than most cable/dsl circuits, no? Josh Luthman Office: 937-552-2340 Direct: 937-552-2343 1100 Wayne St Suite 1337 Troy, OH 45373 On Sun, Aug 9, 2015 at 5:55 PM, George Skorup <[email protected]> wrote: > Yes. Ran out of time and had to get the link back up. Could've been > something else stupid I overlooked though. I'll mess with a pair in the > office next week. > > On 8/9/2015 4:48 PM, Josh Luthman wrote: > > Did you set the slave to eptp slave? They use different drivers. > > Josh Luthman > Office: 937-552-2340 > Direct: 937-552-2343 > 1100 Wayne St > Suite 1337 > Troy, OH 45373 > On Aug 9, 2015 5:41 PM, "George Skorup" <[email protected]> wrote: > >> OK so back to ePMP PTP. I had to put up a 500' link between two grain >> legs. Used integrated radios on 2.4.3. I tried ePTP, but they wouldn't link >> up. Thinking maybe too short? I don't know, so I just switched them to TDD >> PTP w/ flexible framing instead. Latency is like 15ms. I hate it. I have a >> couple other links still on 2.3.3 also doing flexible framing and latency >> is very consistently 7ms. Only thing I can think is that this new short >> link is on 5.4GHz and it's actually forcing fixed framing (even though it >> says flexible) because of DFS? Either that or something is very different >> between 2.3.3 and 2.4.3. >> >> Anyway.. we were thinking about switching out a bunch of 5GHz Rocket >> links (because they cannot handle noise at all) with ePMP GPS. We get very >> good latency with the Rockets, typically under 3ms when the front ends >> aren't swamped, but I will NOT do 15-20ms with ePMP, that's just stupid. >> The whole idea was sync, and since ePTP doesn't sync.. not sure what to do. >> Maybe 2.5ms framing and 50/50 TDD will get me 7ms, and w/ sync? >> > >
