The stablest systems I had the pleasure to manage were two identical rhel6 clusters into different geographical locations for higher availability and fault tole

The stablest systems I had the pleasure to manage were two identical rhel6 clusters into different geographical locations for higher availability and fault tolerance. Such systems were installed, turned on and never touched again. Kernel 2.6.32 that managed about six-seven years of uptime up to mid-2019. We operated a lot onto such systems, mounting and unmounting iscsi devices, starting and stopping stuff, turning on and off network interfaces and clustered filesystem (thanks to Veritas cluster manager). The key move was never updating. Such systems were literally mission critical, without that cluster the whole company was unable to produce its main products. Considering how much stuff they ran and how many simultaneous users were connected, I was humbled by their stability (and by rhel's stability). If you're getting angry at this post: the customer was not in the it field and was completely okay with buying new hardware and doing a full reinstall every X years.