I’m guessing I’m maxxing out the same-unit bandwidth test (WRAP boards) on turbo, I haven’t had a chance to really get a test done that’s using other boards. I have a couple RB 512’s coming in next week I want to try it with, but that’s some awesome throughput for a WRAP board - although I haven’t test like this since 2.8.28… Are these normal sounding numbers, or more because of 2.9 than the new package?
I tried using 2.9.8 with the standard wireless (not wireless-test), on a similar config, both WRAP boards, except at a distance of about 3.5 miles. I was able to get about 18Mbps TCP, 25-28Mbps UDP, with NStreme enabled. It’s an active, important backhaul, so I’m not willing to go to wireless-test yet, but whatever they did in Wireless-Test, it is a lot more efficient than the standard 2.9.x wireless.
Some simple steps will do, but there ARE some caveats
Insert the CF card into a card reader in your Windows PC.
Install MikroTik RouterOS onto that CF card using the netinstall program.
DO INCLUDE the routerboard package! (Otherwise RouterOS will NOT really reboot later if you issue a reboot command!)
Put the CF card into the WRAP board, boot up. After RouterOS finished installing itself it is trying to reboot - but won’t (!) succeed this first time. You have to manually reboot by removing power to the WRAP board for some seconds here.
After that you have a running RouterOS install on the WRAP board.
Most people telling they couldn’t get RouterOS running on a WRAP board tend to have problems with the serial port settings. There are two “things” taking control over the serial port on a WRAP board when booting:
The WRAP BIOS during the first boot stage. This has a default baud rate of 38.400 bit/s.
The RouterOS serial console (after RouterOS has booted). This has a default baud rate of 9.600 bit/s.
So you either have to
switch baud rate in your terminal program at the right time or
make both parts use the same baud rate (more sensible).
You can change the serial port baud rate in RouterOS, but when issuing a “system reset” command (thereby erasing all config in it), it’s revert to 9.600 bit/s.
So my suggestion is to set the baud rate of the WRAP BIOS to 9.600 bit/s, too. If you only use the serial console for first config steps (activating interfaces, adding ip addresses) and do the rest via ssh/WinBox, this won’t hurt much…
Best regards,
Christian Meis
PS: Yes, I know - this should go into the Wiki… If time permits…
Those #'s were very similar with and without NStreme. I posted another one with both sets of #'s on the Wireless forum. I have a set of RB512’s now that I’m baselining with the 2.9.8 regular package first, and in a few days I’m going to run external speed tests on it.
Have you tried this process with any of the 2.9 releases? I cannot get past “Starting services…” when installing 2.9.10 on a WRAP but with 2.8.28 there are no problems at all. I am woundering if it is to do with the brand of flash card (biwin) but unfortunately I have no other brands here to try.
Last time I did this with 2.9.10 only some hours ago. This is on a WRAP.2C (the smaller one, 1 ether, 2 MiniPCI.
Everything works normally and as I described (if you take the required manual reboot during install as normal, that is)…
FYI The problem I was having trying to run 2.9.10 on a WRAP.2C was some how related to the console cable I was using. Tried a different cable and worked fine. swapped cable back and again could not get a login prompt.
I am seeing similar fast speeds on the WRAP platform. Is there are technical reason why there is such an improvement? I heard there was a letency issue with the old driver that was resolved in this one. Is that the reason for the improvement?
It becomes a regular package