I've got a main ethereum node for business use running Parity version
v2.5.13-stable-253ff3f-20191231/x86_64-linux-gnu/rustc1.40.0 on an AWS instance and a near exact replica of that setup as a "clone" backup in case something happens to the main node. Both are connected and live.
The only difference is that the backup ran out of space a month ago and I had to re-build the database after upping the disk capacity. I then installed Parity
v2.7.2-stable-2662d19-20200206/x86_64-unknown-linux-gnu/rustc1.41.0 instead of the previous
v2.5. For some reason, it refused to warp sync, and so I let it just slow-sync as it wanted to. It took a few days, but got back up to speed and has been synced for several weeks now.
What I can't figure out is that, despite being set up identically, the two nodes have vastly different disk usage. The main node is currently using 289G, while the backup node is using 321G. Furthermore, when you look at the munin disk usage graphs of the two during the exact same period, the difference is remarkable:
In particular, the rhythmic oscillation of the clone node seems concerning. However, I can't find anything in the logs that would suggest they're acting any differently. When I follow the logs of the two nodes, they're almost identical (as they should be).
Any thoughts are appreciated.