HPE 2930M MPS Absent warnings

Executive summary: the Right Way apparently doesn't work, the trunk way seems to work. What to do?

I'm new with HPE switches so when reviewing the documentation I skipped over HP's notion of trunking since it looked like bonding to me and not what Cisco calls 802.1Q trunks, which is what I want.

Searching this subreddit I discovered that I don't need to define a trk between switches, just tag the VLANs that should transit among the 3 switches and be done with it.

Yet, I had done exactly that, tagging the necessary VLANs originally on my SFP ports between the switches and noted later that I had a majority of my PoE endpoints dropping off the network with "PD MPS Absent" indications. Everything worked fine individually during configuration. All of the runs are less than 100 feet of copper, so I dismissed a physical connectivity problem. The power budget for each switch has more than enough power supply current to serve all PoE endpoints simultaneously.

In a panic to get this network ready I created an HP-style trunk between 2 of the switches as a test and everything settled down, so I figured I had struck upon the cure. But upon further review it seems that goes against recommendations here. Quandary.

What's the best practice in this case? I can't understand the cause of the warnings when the inter-switch links were configured with only tagged VLANs, but the trouble disappeared when I used the HP trunk command?

This simplified network diagram shows the topology and cogent configuration of the core switch when I had everything configured to tag VLANs on the SFP ports, when endpoints were dropping off line mysteriously.

Thanks for guidance and best practice in this case.