I have a few CentOS 5.3 servers out at remote sites that connect over a satellite link (a painfully slow satellite link). There's a time period in the wee hours of the morning when the satellite provider doesn't stiff me for bandwidth, and when the link is generally idle, where I want to use rsync to back up files to my local server. I've got rsync working in in combination with cron and ssh, and that process works fine.
The problem is getting the initial copy of the files over to the local server so that rsync can just deal with files changed during the day (there tend to not be too many changes -- mostly office-type documents, few pictures, that sort of thing, so limited bandwidth is not a problem). If I run rsync now, however, it is trying to copy over all the files since none exist on the local servers, and by my calculations copying that 160gb worth of data should be done some time in mid-February, 2010.
I can make the 4-hour round trip out to the remote site with a USB hard drive and copy the files that way, but my concern is how to get the file structure, timestamps, permissions, etc copied exactly from the remote server > USB drive > local server so that rsync can run most efficiently (and not have to do a compare on every single file the first time it runs.)
So the question is, how do I copy the files from the remote server back to my local server in a way that will allow rsync, once the copy is complete, to start dealing just with the daily changes in the most efficient way possible?
Thanks!