Silent Corruption?

While comparing the contents of my Drobo (version 2) with its backup, I noticed that of the ~13,000 files, about 15 were different. In each case (where I could determine it) the file on the backup was the good one and the file on Drobo was bad.

Is this a common issue?

I was using rsync (with -cnav) on my Mac Pro (Leopard 5.8) to compare the two directories. Drobo was connected via FW800 and the backup Disk uses eSATA.

The backup disk is rarely on (only when backing up), but Drobo is on all the time and is usually connected to my Time Capsule and shared to the rest of the network.


hmm, its not something i have come across, but the majority of stuff on my drobo is video files so i may not notice the odd bit here or there, plus for the majority of it, the drobo copy is the only copy, so i dont have a known good to compare to!

Definitely not a common issue with me. I’m using ViceVersa Pro (for Windows) to copydata with full CRC verification after copy (and a few more times since I’ve restarted a few comparisons) and so far 2 TB of data has copied successfully to my Drobo.

First thing is run a filesystem check. Reboot, then I’d do is grab an MD5 or SHA1 hash generating utility and get the hashes for both files.

Generate some huge files, copy them around your non-Drobo drive(s) first, then copy them around your Drobo. Compare all the hashes.

Way back in the day I saw weird network corruption and caching issues back in Windows 2000, and that was a Windows networking (oplocks) issue. You’re on Mac OS, so it’s difficult to say where the issue arises here. Just want to point out that the OS is not out of the question either, which is why I suggest making copies around the non-Drobo drives too.

Also check for malware.

Check your cables - both the FireWire800 to the Drobo and the eSATA to your backup disk. Make sure they are not wound around any power cables.

Or running too near TVs, VCRs, computer chassis and power transformers (aka “bricks”). Cables are supposed to be shielded from such things, but in a typical environment you’d be surprised how much electrical noise there is. It definitely doesn’t hurt to route signal cables away from power cables.

I have seen files becoming corrupted when I copied them from one location to another or if I worked with them. It turned out to be a bad RAM with a very sneaky error that passed most of the tests. Did you copy or move those files?

I used SFV-checksums to verify my files. There are loads of apps that support SFV.

Have Fun

Good point STB. Running a good memory tester like memtest86+ through at least 3 passes would be highly recommended.

I’ve had numerous occasions where one stick or set of RAM that is perfectly fine on its own does not play nicely with another stick or set. RAM is a very finicky thing, despite what the memory manufacturers like to tell you.