Frequently Asked Question

rsync: versatile data copy
Last Updated 2 years ago

When it comes to copying, updating, or in some other way moving your data around, there really is little that can compare to rsync.

As was discussed during the Aeolus batch/training seminars, we suggest using rsync to upload and manage your data within Aeolus. cp and mv don't always maintain permissions, attributes, file stability, et cetera while moving data across filesystem types or servers (especially across the network). If you are not satisfied with "cp -au", we suggest again rsync. In this tutorial, we will also be covering the use of screen a lite terminal emulator and tail a terminal printer. When in doubt, use/check the man page(s).

Learn about the Commands

Consider the following rsync options:
  • h : output numbers in human readable format
  • a : archive or "-rlptgoD" for recursion, permissions, etc
  • i : output a change summary of all changes
  • v : increase verbosity
  • z : compress file data during the transfer
  • --dry-run : perform a trial run with no changes made
  • --progress : print information showing progress of transfer
Consider the following screen options:
  • S : when creating a new session, specify a name
  • R : attempt to resume first detached session it sees
  • ls : prints a list of screen sessions
Consider the following tail options:
  • f : output appended data as the file grows

Run the Commands

The basic rsync command holds the following syntax:
rsync  [from] [to]
rsync  [source] [target]

Putting that all together, you should be able to do something along the lines of:
[aeolus #]: screen -S syncdata
[aeolus #]: mkdir ~/logs
[aeolus #]: rsync -haiv --progress /fastscratch/myDIR/mysubdir /data/myLAB/users/myDIR/ > ~/logs/syncdata.log
* NOTE: trailing forward slash "/" will grab data in a directory or move data into a directory
* NOTE: no trailing forward slash "/" will grab the directory or write on top of a directory

By typing CTRL + "a", then separately "d" to detach the screen session, you will be returned to your previous shell.

To watch the progress of the data copy:
[aeolus #]: tail -f ~/logs/syncdata.log

Finish with the Commands

By typing CTRL + "c", you can kill the appending tail stream/view.

To list out active screen sessions, you can type the following:
[aeolus #]: screen -ls

Then to re-attach your previous screen session:
[aeolus #]: screen -R syncdata

Once in, by typing CTRL + "c", you will kill the current re-attached screen session.

* NOTE: It will be up to you to check your log file for errors. Configured as above, rsync will not be able to copy files with read-only permissions (as it has to be written somewhere).

Bonus rsync Commands

* WARNING: These are to only be exercised with extreme caution. If you get the paths wrong, rsync will do what you said and likely purge unintended data (that you have access to).

If the data that you were copying has been updated (more or less written to it) since you started the copy or your last copy, that's not an issue! You can just run an rsync as described above and then manually handle duplicate data, or you can use --delete-delay in your rsync command.
  • --delete-delay : find data that no longer exists in source and delete it from the target after the transfer
[aeolus #]: rsync -haiv --delete-delay --progress /fastscratch/myDIR/mysubdir /data/myLAB/users/myDIR/ > ~/logs/syncdata.log

Additionally, you can remove the data from the source, similarly to "mv" but better.
  • --remove-source-files : sender removes synchronized files that are part of the transfer
[aeolus #]: rsync -haiv --remove-source-files --progress /fastscratch/myDIR/mysubdir /data/myLAB/users/myDIR/ > ~/logs/syncdata.log

Please email us if you have any questions or issues.

Please Wait!

Please wait... it will take a second!