When a Crashplan backup seems far larger than the original data – finding junctions
I recently had an issue whilst setting up a PC on Crashplan, where it never seemed to stop scanning the source files, and kept adding more and more files!
I checked the size of the two source directories, and found 12,000 files in each for a total of 8GB – yet Crashplan had counted 80,000+ files totaling 35GB, and was still counting and counting….
I suspected some kind of loop, and that a Windows shortcut in the directory was to blame. For example, I had a shortcut to “My Documents” inside the My Documents folder, and my thought was that Crashplan was recursing through this. However, it should be able to handle Windows links without following them, since there are often plenty of symbolic links in any users’ directory.
After some investigation, I realised that the link was not a normal link… but a junction. Junctions are like links, but ‘harder’ – they are indistinguishable from a genuine directory by normal applications, which is why they work as a way of redirecting iTunes to a backup directory on another drive, for example. However, if recursive, they’ll also confuse the heck out of backup software, which is why I found that the backup contained D:\My Documents\My Documents\My Documents\My Documents\My Documents\file.doc. The fact that it looks like a normal shortcut in Windows Explorer…
….such as someone might accidentally create by fat-fingering a mouse drag, meant I didn’t spot it.
You can check for junctions using the command:
dir /aL /s C:\
Typing this will show a list of all the junctions it finds, recursively. You can then remove them – I actually just deleted the link in Windows Explorer.
After this, I checked again… and voila – Crashplan counted the right number of files.