Syncing on gaia-7001 (graphs!)

Some users noticed syncing new nodes on gaia-7001 was slow, and getting slower over time, so I ran some tests and generated some graphs. To generate the data for these, I ran commands like: `gaiad start --pruning=x | grep "Executed block" | tee x-machine_size.log` I then used the following script to get the difference in timestamps between successive blocks from the log. ``` import time fname = "everything-micro" l = open(fname+".log","r").readlines() ts = [li[2:20] for li in l] tsf = [time.mktime(time.strptime("2018-"+t[:-4], "%Y-%m-%d|%H:%M:%S"))+(int(t[-3:])/1000) for t in ts] diffs = [tsf[i+1]-tsf[i] for i in range(len(tsf)-1)] open(fname+"-diff.txt","w").write("\n".join([str(d) for d in diffs])) ``` To generate the graphs I used gnuplot, and simple commands like: `plot "everything-micro-diff.txt"` First test I ran was with pruning=everything on a t2-micro box on AWS (1core 1gb of ram). By default nodes are pruning most old states, only keeping one every 10,000 and the last 100. Since the network was below block 10,000, most nodes would just be keeping the last 100. By setting pruning=everything, the node would delete historical states as soon as a new one was…

Excerpt (1194 of 3733 characters). Read the whole post on the forum ↗