It's been a wild year, and certainly one that will go down in history for
So, having monitored the node overnight it seems high CPU, low ram and large swap is not suitable going forward. I expect the speed of the swap and extra overhead of writing to the swap is what caused the node to skip on updates quite frequently, as seen below.
So I think in fact the opposite here is what we should be aiming for. It seems plenty of RAM is critical for this application, and CPU much less so. You can actually see in the graph below how GOD pool performed at epoch transition, with just a single CPU. It never even hit 100% so it's clearly not the bottleneck here.
Just now I've upped our RAM and lowered the CPU, and I think this might be the recipe I'm after to keep things lean, whilst also being truly optimal and performant. We don't want to ever miss a block from an underpowered node, and conversely I don't want to be paying for resources that are sitting idle. This is all about finding the sweet spot. I'll monitor it today but it'll be the epoch transition where it'll be put through its paces. That's on Saturday 8th August.