How *not* monitoring your backups can cause data loss

talmy

macrumors 601
Original poster
Oct 26, 2009
4,713
275
Oregon
It is important to test your backup plan to make sure it can actually restore files, and it is also important to monitor that backups are actually occurring. Watch for error messages and read any backup reports for abnormalities.

I didn't lose anything when a sector on the boot drive on my Mac mini server failed, but I sure could have. Here's why one little failure in a seemingly unimportant file can cause a complete loss of backups. It's much like "for want of a nail the kingdom was lost..."

I back up the two important (critical changing data), internal drives on the server automatically, nightly using SuperDuper! to another attached drive. It's part of a pair of external drives I alternate and keep one off-site. I also clone the external drives with infrequently changing data manually. Finally I use CrashPlan to continuously back up off-site about 1.1TB of data spread across several drives. Redundancy is very important.

A sector failed on the boot drive (the drive is 4.5 years old and runs 24/7 so should I be surprised?). The SuperDuper! program indicated failure to back up and the log file showed the problem. Note that when this happens the backup operation quits leaving the backup drive in a potentially indeterminate (i.e. useless) state. Luckily the program allows selecting files to not back up, so I selected that file allowing the backup to continue. I've ordered a replacement drive, but because of travel there will be a 2½ week lapse between discovering the problem and replacing the drive.

But this wasn't all! That "nail" caused another backup problem. The file in question was part of Crashplan's cache. CrashPlan appeared to work but I noticed the report email from crashplan.com was stating 120GB was being backed up and not 1.1TB. Data on four different drives was not being backed up because of a single failure on one drive that just happened to be in a critical (for CrashPlan) location! Luckily the CrashPlan website gave instructions for moving the location of its cache files, which I did to another drive, and got it running again. Of course I had to leave the bad file in place so the bad sector wouldn't get reused. But, oops, the bad file has a new name which caused SuperDuper! to fail again!

Anyway it's now running until I get that replacement drive. Note that even if I get a full drive failure while I wait I still have at least one good cloned backup boot drive I can run from.

Never underestimate the need for backups, and make sure they are actually being made!
 

talmy

macrumors 601
Original poster
Oct 26, 2009
4,713
275
Oregon
Just following up. Replacement drive arrived and it turned out that my boot drive had deteriorated some more. Of course I had stopped backing it up (so as not to corrupt the back up drive), so I cloned the replacement from the backup drive.

Now it's all running again. The failed drive was one of the original drives so I guess it had every right to fail.
 

Traverse

macrumors 604
Mar 11, 2013
6,869
3,101
Here
I've even been burned by Time Machine. You need more than one back up of critical files.
 

talmy

macrumors 601
Original poster
Oct 26, 2009
4,713
275
Oregon
I've even been burned by Time Machine. You need more than one back up of critical files.
IMHO TimeMachine is only good for recovering from accidentally deletion or modification of files. I've had it fail far to many times to rely on it for system backups.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.