It’s week 9, which means that you’ve probably invested many hours in a final paper or project for at least one class.
This is your reminder to stop and ask yourself:
If the most important file in my project suddenly disappeared, what would I do?
If the answer is
git checkout -- filename or
cp filename_yesterdays_date filename, you’re on the right track! If losing the file would ruin your day,
take a few minutes now to prevent it from getting lost in the future.
Verify your backups.
If you think that you know how to restore your file, do a test right now. Make a copy of the project’s entire folder and, in that copy, actually delete the file. Now try to get it back. If you don’t have a plan for getting your work back, or the backup solution you thought you had isn’t working, you can set one up now while your work still exists!
Decide how serious a disaster you want to recover from.
You might face one or more of the following threats:
User error, such as
mv junk importantfile,
tar importantfile junk, or
cat junk > importantfile
Hardware failure, such as dropping your laptop and irreparably breaking its hard drive
Local catastrophe, such as your backpack getting stolen (containing your phone, laptop, and USB stick)
Regional disaster, such as an earthquake (your laptop, USB drives, and Flip are all destroyed)
The secret to avoiding file loss in all of those scenarios is to have multiple off-site backups. In the situations that you’re likely to encounter, having even one recent backup can save you hours of work re-doing your project.
Which files should you back up?
One rule of thumb is any file that you modify by hand. Files that only get modified by your tools, such as executables that you’ve compiled, can be regenerated whenever you want as long as their sources are available.
If you generate executables from c code and you have the .c files and Makefile safely backed up and verified, there’s no need to back up the executables as well.
If you generate a pdf from LaTeX source, you don’t need to back up the pdf as long as you have the .tex file backed up.
Here are some techniques for backing up your work:
Email a copy to yourself every night. If you use gmail, even the destruction of every computer in Corvallis wouldn’t cause you to lose your work, since Google stores copies of your mail in datacenters all over the world.
Keep your code in a repo on GitHub (which offers 5 free private repositories to anyone with a .edu email address; see https://education.github.com/) or BitBucket (which offers unlimited free private repositories but is a less effective portfolio of your public work). Commit every change you make, and push to the remote repository frequently.
Make a copy with a name like
myfile_description_of_changesevery time you make a major change, and save that copy on a USB stick (or even a different folder on your computer)
Transfer your files to flip.engr.oregonstate.edu, using a program like sftp or FileZilla
Use an rsync tool, such as obnam or rdiff-backup, to copy only the changed parts of your project onto an external disk or into a different folder. You can run it by hand every time you want to save a change, or set up a cron job to run it automatically at specified intervals.
When should you back up your work?
Every time you make progress on a part of the project, however tiny, and think “wow, I never want to have to do that again”.
Every time you think “I might want to put it back like this if the change I’m trying doesn’t work”.
Before deleting any part of it, for any reason
Before running a shell command that could inadvertantly delete or overwrite it. Some common culprits are
>>) and any command to your compiler that specifies the input filename as the output file.
Please take a couple of minutes right now to back up any work that it would be inconvenient to lose. If you have any questions about open-source version control or backup tools, or need advice on finding the one that’s best for your use case, just ask in #osu-lug on irc.freenode.net or on the mailing list!