I'm not sure this is leading to anything, but I tried turning off aggregation.
R3 -> R4 UBNT link throughput test dropped to about 30 Mbps (less than half).... but my R1 -> R4 test almost doubled to about 20 Mbps. [I'm using firmware v5.6.9, but I think this issue has been present for a while.] I experimented with various frame / size settings, but did not see any further improved results. I tried enabling flow control on the MikroTik ports used for the UBNT link, but that did not help (and no pause frames were recorded). It is probably time to take a break and mess with it tomorrow. Thank you for the help today as usual. On Wed, Oct 26, 2016 at 7:15 PM, Christopher Gray <cg...@grayred.com> wrote: > R1 is connected to the Internet via an MRV auto negotiated to 1G FD (This > is a 1 G Internet connection) > > R2 is connected to the Internet through a copper connection to a Juniper > switch on a 30 Mbps layer 2 fiber transport to a switch where the Internet > service is connected. This R2 -- Juniper connection is set manually because > they would not play nice if set to auto. > > I have a similar setup on the same fiber system in 3 other locations. None > of which have this behavior (but also, none of which have UBNT M5 hardware, > so that could be a potential issue). > > > On Wed, Oct 26, 2016 at 5:31 PM, That One Guy /sarcasm < > thatoneguyst...@gmail.com> wrote: > >> define your transport, are you talking a layer 2 circuit, vpn tunnel, >> fiber, etc? >> >> On Wed, Oct 26, 2016 at 4:24 PM, George Skorup <geo...@cbcast.com> wrote: >> >>> The manual 100FD interface... what is that talking to? The Auto 1G on >>> R1? If that's the case, I'd bet that's your problem. Keep in mind that you >>> cannot run auto on one side and fixed FDX on the other side. This results >>> in a duplex mismatch. The interface in auto will fall back to HDX. If you >>> did auto one side and HDX on the other side, they'd both be HDX, so it >>> would work fine. But obviously half duplex sux. >>> >>> On 10/26/2016 1:54 PM, Christopher Gray wrote: >>> >>> R1 is the only router with 1 Gbps ports. Everything is auto except 1 >>> connection that requires manual settings. >>> >>> *R1* -- (Auto 1 G FD) ...Internet... (Manual 100 FD) -- *R2 *-- (Auto >>> 100 FD) -- *R3* -- (Auto 100 FD) ...M5... (Auto 100 FD) -- *R4* >>> >>> MTU is set to 1500 on every port (and the UBNT link). >>> >>> Flow control is off, and none of the interfaces show any pause frames >>> received. >>> >>> >>> This is a live link, but it is only running ~ 1 Mbps otherwise. >>> >>> >>> >>> On Wed, Oct 26, 2016 at 2:12 PM, Steve <li...@wavedirect.org> wrote: >>> >>>> Few questions come to mind. >>>> >>>> Are all set to auto negotiate or are they fixed at 100Mbit? >>>> What are the MTU's of each connection? >>>> Flow control turned on? >>>> >>>> >>>> ----- Original Message ----- >>>> From: "Christopher Gray" <cg...@graytechsoftware.com> >>>> To: "af" <af@afmug.com> >>>> Sent: Wednesday, October 26, 2016 1:57:56 PM >>>> Subject: [AFMUG] Trouble Identifying Throughput Issue >>>> >>>> I have a section of my network that is lacking something, and I can't >>>> figure out where the problem is. I'm looking for any thoughts / >>>> suggestions. >>>> >>>> 4x MikroTik routers >>>> >>>> Link speeds: >>>> R1 -- (30 Mbps IP / Transport) -- R2 -- (100 Mbps Eth) -- R3 -- (75 Mbps >>>> UBNT M5) -- R4 >>>> >>>> The limiting factor for traffic should be the Transport, and I expect >>>> to be >>>> able to get 30 Mbps across the system (one-way). >>>> >>>> Testing from R1 to R3 runs 30 Mbps. >>>> >>>> Testing from R2 to R4 runs 75 Mbps. >>>> >>>> Testing from R1 to R4 only runs 10 Mbps (instead of 30). >>>> >>>> Tests were one-way btest with 20 TCP streams. >>>> >>>> Any ideas for something that would cause this? >>>> >>> >>> >>> >> >> >> -- >> If you only see yourself as part of the team but you don't see your team >> as part of yourself you have already failed as part of the team. >> > >