can confirm PRECISELY this behaviour (and more) -- there are major issues with this product's WiFi stability and performance at the moment
1. sometimes wired backhaul creation fails to form mesh properly (startup race condition? islanded mesh connections?) <-- especially bad if you get a power blip/outage ... you have to retry powercycle sequence a few times until the network actually comes online properly
2. clients can get "stuck" into 2.4GHz band (steering stops working) -- even when next to secondary node, refuses to flip to 5GHz
3. when on 2.4GHz speeds are horrendous (~1-20 Mbps)
4. when on 5.0GHz speeds are fantastic (~200-500 Mbps)
TIPS:
I used `iperf` tool locally to eliminate ISP/Internet variability which was drastically helpful, as well as took packet captures of the D-Link topology traffic in hopes that support/R&D can isolate the problem.
I used `wavemon` and `linSSID` on a Linux notebook that supported by 2.4 and 5.0 GHz bands to analyze -- confirmed wifi signals/channels are clean and this is not relevant to the problem.
Confirmed with D-Link support that when the COVR points are in bridged mode w/ wired backhaul, the placement really shouldn't matter or affect performance. (I saw several points about "at least 30' apart" ... but not too far apart, etc -- this is not relevant when using wired backhaul and placement should be flexible since WiFi mesh nodes will auto-select their channels to avoid conflicting with one another)
I contacted D-Link support and was confirmed that the case is escalated and under lab testing.
Case ID: `C6984446`