I have a 2TB directory that is 99% the same on the remote system. Once every few months, I do an rsync to the remote machine/directory with --checksum just to make sure there's no bitrot. I did it recently after a RAID verify failed, for example.
Here's the issue: it takes 10+ hours to run, but it appears to wait a really long time to start the checksumming on the remote system. It appears to do it first on the local system, then remote, then compares them and then copies any files that are different/new.
It seems to me this 10+ hour process would be cut in half by doing the checksums at the same time on the local and remote system. Yes?
I'm using rsync 3.0.9 on CentOS 7.1. I did look to see if newer rsync's have this feature and I didn't see anything, but maybe I missed it.