The mining hardware has been rock stable, i made a huge adjustment in the way the mining is implemented on my rigs, previously i had all the rigs setup farming as group of cards and after talking with someone who is a much larger farmer than im currently at, he experimented with solo-card farming and found it made his rigs more stable and made troubling cards stick out far easier, allowing for diagnosing to be easier. I ungrouped all my rigs and now have the GPU's mining solo, with their own specific benchmark settings to determine which algo is best for it to switch to. So a 6 card rig has 6 individual miners doing their own thing, mining their own algo's based on individual benchmark settings. Some huge takeaways from this, is that the rigs that i was having random lockups/hiccups on, the rig itself no longer goes down as a whole with miners not loading due to cards greyed out or frozen, now only the GPU on those rigs that is causing the headaches is locking up and the rest is just happily farming away. I also installed a TV in my office that runs off an amazon firestick, which is web browsed to my awesome miner display via the local host, i have it setup on the simple dashboard that only shows the amount of GPU's mining and estimation of profits, so if i look up at that screen i can quickly see if a GPU is down from the number changing... then i look into which GPU it is and proceed from there...
With each GPU running a separate miner, have you had any issues with the CPU (or RAM) not keeping up? In our experience the system gets too laggy and unstable if there's 3+ miners open at the same time. That would be with cheap dual core Celeron/Pentium CPUs and 4GB RAM; on our Ryzen 5 1600 rigs the CPU has no problem keeping up of course.