I found a lot of the videos from people claiming to have the straight forward way of setting up mikrotik routers are great sounding, confident, but do it in a convoluted way and messes up the overall configuration of the router. They are the people who have a small understanding and just making videos to make money, and some of these people are the ones who charge a few dollars to configure your router for you. Then you find the router doesn’t work very well.
Having experience in routers I gave up on those videos and found a straight forward one and the setup is really quite simple once you know.
For now, I’m routing old Fritz!Box 7530 wifi traffic through ethernet on the Mikrotik router, so I suspect my wifi channels are a bit crazy. Once I’ve migrated everything over, I should get a better idea about performance. I’ll setup those port maps tomorrow too. Impressed so far though (see here: Upload tests - #15 by Traktion)
Thanks. I have similar setup as you have, port forwarding done through the NAT tab. I see that you also have simlilar as me a significant amount of data “not coming from Lan” that is being dropped.
I had a previous problem with having forwarding rules both in firewall and NAT tab which gave that 25% of packages were dropped, that one was fixed. Still the Hetzner nodes seem to earn more and faster every time than the home nodes, trying to figure out why,
I also learned that the bucket size in queues if set to example default 0.1 value then that represent 1MB. So I changed that value to 5 to allow for 50MB, after that I have not seen any dropped packages. Memory usage on the router is around 500-600MB so there seem to be good headroom up to 1GB max.
Yes and its a simple rule. Mine is set to around 37Mbps up
But you have to realise that the PC still pumps out the data as fast as it can and the router has to buffer. So no free lunch, only good to smooth out peaks which it is good at. My buffer is approx 100MB (800Mbits) which ends up being 20 seconds of buffer. That is really too much
@Shu I have discovered that machines with higher latency earn significantly less, hold less records and less live records despite not being shunned.
Took a bit of head scratching to figure this out because on the face of it all was well, peers, records, no shuns, yet horrific earnings.
Had I not had adjacent machines to compare these too, I would never have noticed that there was a discrepancy.
I would have thought that is just what it should be like.
The culprit machines are on a long ethernet cable that slowed them down.
The cable is really long, 40’ in my case. The issue though is a poor short cable may have the same affect and for home node runners that can’t compare several machines they may never know that they are hamstrung by something stupid like that.
No it was all a bit of a rushed discovery this morning before work and interrupted by the cat bringing a rat in followed by my wife insisting I tear the house apart before work or nodes. Will need to wait until later.
Its error rate in your cable, not the length. Your internet connection is so long in comparison that it dwarfs that tiny distance. Your comparison if length was the issue would show exactly the same.
More like retries means that sometimes your quotes do not reach the quote requester. That’ll be the much more likely reason.
(I used to work in this space for a bit…)
What!? Cable length really matters that much? How long the cable?
