Love this kind of information. Learning something new everyday. So plugging the NAS into a 66U close to the Vero box should help?
I just enabled link aggregation on the NAS and on the router. If this doesn’t fix anything, i’ll un-bond the connections and move the nas closer to the Vero box.
Let’s not to forget, the issue appeared only with the Vero. All the things we ask you is to sort out possible issues with the network setup, so setup LAG and move NAS will possible help you reduce the issues.
But:
Did you setup the connection on Vero towards your NAS the same as on your mimix?
Do you still NFS fstab mounts to your NAS?
Do you resolve names for connecting to NAS or just via IP?
You mentioned running another test now after setting LAG on. Did the issue you’ve had still reappear?
Oh, sorry. I must not have made that clear enough. It also surfaced on a Minix UH9, which I replaced with the Vero. And a RPI2B struggles with playback as well. Don’t know if that’s related though. I use it too little to really want to dig into this problem.
Yes, sir
IP only.
it was not clear to me 
about the other devices showed the same issues. Then the only difference now is Kodi 18.
So then we are on the “right” track with network investigation as all devices having the same issues. Could also be related to certain “sources” as @dillthedog mentioned but it’s better to clear the network from the unexpected retries as shown into the iperf logs.
Thanks for the reminder. I will run iperf3 again now that I’ve disabled all the family stuff on my router and enabled link aggregation on the NAS. Will do so right after another test run with a movie featuring a lot of Galileo’s.
Forgot to answer this one:
Because I started out with cat6 during a renovation when I had the opportunity open up the ceiling and place some cable there. Once the renovation was finished I realized I pulled the cable too hard. Now it will only support 100Mbit. That 'll teach me.
There is no way to really answer this other than with a ‘maybe’. In theory as long as everything is working as it should and your total bandwidth is below 1Gb then it shouldn’t make a difference. Your route from the NAS currently running through a switch that has the ability to affect its traffic, and has all the torrent, netflix, whatever else flowing though it at the same time. In addition once it leaves that router it is running through three more switches and their cabling so it gets to share every point of failure and all traffic going on in the house. If it was sitting on the other end of your network then your traffic goes from the NAS into a switch and right back out to your player and there is not a lot to go wrong there. You could even go one better and move your device with the torrent client to the main router to lower the overall traffic even more. I wouldn’t really worry about losing LAG by moving the NAS as it doesn’t actually get you anything the way your house is wired.
Thanks, I’m going to go up to the attic this weekend. I killed a few processes on the main router. It helped. Two problematic movies played right to the end. Even with another 4K movie playing on a Minix 1.
A third movie on the Vero box however, quit somewhere around 2/3. Don’t know if this was my fault though, I was fiddling around with some network settings. I rebooted the NAS just to be sure.
I enabled LAG on both NAS and router. Osmc responded by staying black on boot. So now I’m back on load-balancing on the NAS.
Will report back as soon as I moved the NAS.
If the NAS wasn’t plugged into the router then it wouldn’t lose its connection when you changed settings on your router. 
Which is what happens when it can’t access a MySQL database. You would need to check to see if enabling that changed the IP address it was using. Probably not worth messing with as it would only have benefit if all your devices were running directly into that router. As the rest of the house is running in/out from a single line you can’t really take advantage of LAG.
It was plugged in. Something else is still going on. I’m still seeing lots of retries every first run of iperf3. It’s like the NAS is asleep, which it cannot be. Disks are set to sleep only after three hours of inactivity.
I meant that when you change settings on a router it often has to reboot or at least reinitialize some of the hardware (like the managed switch) which affects whats plugged into it. What I meant was if the NAS was sitting on the other end of your network it would not be impacted by anything going on with the router.
iperf does not use the hard drives.
Keep providing me with reasons to move the NAS. You’re building a rock solid case for it so far
Seriously, your advice is highly appreciated. 
Bring the vero next to the nas at ground floor next to main router and login with ssh and just check iperf only between vero and nas. So no display to move. That will save a couple of routers in between and a switch. Just step by step sorting out what causes the issues.
That sounds like your problem. I bet the problems are all related to that cable. If you pulled hard enough to break some of the wires, you probably have a marginal connection on the others.
This cable is no longer in use. I created a new connection from a router in the attic using cat6 that I manage not to damage this time. I could add it to the switch that the Vero box connect too as well, but I see little added value in doing so.
I know it has already been suggested, but have you confirmed that the problematic files play ok on local storage? Apologies if I missed it, but I can’t see an answer to this basic question.
Looking at the original iperf figures, I see a few bursts of retries, where the bitrate and TCP congestion window both drop slightly, but the network throughput continues to be much bigger than any movie would possibly require and the Vero4K’s caching should be able to smooth over brief glitches such as we see on the iperf stats.
You didn’t miss it, but I tried. I have tried local storage, but not long enough to prove anything. It messes up the centralized storage system that I use for all mediaplayers around the house. I’m running a new test tomorrow using a top HDMI cable. So far I’ve been able to get two 4K movies to play entirely. Only to see a third exiting with messages like “CAMLCodec::AddData: packet to big:” . Good to know: I restored my Vero box. Not by moving the .kodi directory, but by using a prepared SD card.
I did so because I saw a lot of skin- and add-on related errors in my logs.
Update. Fingers crossed I’ve seemed to have found a way to get it to work. And it seems like the Asus RT-88U (a rather expensive and at least partially disappointing) router is to blame. Behold my network layout, including the previous NAS location and the new one.
I tried placing the NAS in my living room, directly attached to the same switch as the Vero box. But in spite of precautions I took to block the noise, the constant whirring and harddrive noises drove me up the wall.
Since I’ve set it up like this, i’ve played 1 hd and 5 UHD movies with Atmos sounddtracks and in one case I simultaneously played a series episode on another Coreelec Arm device attached to the first switch. All without a single error message in the logs.
Iperf3 consistently shows results with bitrates such as 944 Mbits/sec and 0 retries.
Thanks for @darwindesign for pointing me towards my routers as a potential source of drama. And thanks to everyone else for the tips, tricks and patience.
Let’s hope it keeps working. Networks do work in mysterious ways.
