August 8, 2020

Resource Optimisation (Part 4)

Resource Optimisation (Part 4)

Well, epoch transition was a telling one, and I got all the information I needed from it. That's nice because I'll be going into the next epoch 211 (the one where we start making blocks) with rock steady nodes.

If you look at the graphs above you can see how the block height stalls for a short while, then catches up. I had a look at IOHK's nodes to see if that's normal, and sure enough they plateau as well.

IOHK Node

As for relay-1, you can see how it died and couldn't get back on its feet, that's because it didn't have enough memory. I wanted to run this configuration because my goal is to run the nodes as lean as possible whilst not effecting performance. If it had passed the test I would've ran this config for my relays.

Shortly before the epoch transition I upgraded relay-2 to have 2CPU 4GB RAM, rather than the 1CPU 3GB RAM that relay-1 had. That way I had two configurations to compare and draw my conclusions on.

Sure enough, the higer spec'd relay survived the transition, meaning the block producers GOD/Linux could use that relay to get their block updates from.

Block Producers

Having dialled everything in we're set to go! That should conclude resource optimisation for now, I'll take this setup into the next epoch where I expect we won't have any issues 👍.