Another quick little demo showing the self healing feature with ZFS.

In this demo you are quickly taken through how ZFS can recover from data corruption in mirrored volumes. The demo first starts by creating a storage pool comprised of two disks in a mirror, they take a quick look at that mirrored pair health. The demo then proceeds to corrupt the mirrored pair by writing random data to one of the disks to simulate massive data corruption.

Now what I thought was cool, they then checked the health of the mirrored pair and everything was still in a healthy state. This was because everything was stil in memory cache so the system didn't notice the corruption on disk.

The demo then forced a flush of the cache. To my surprise, when they looked again at the health of the mirrored pair, there was no corruption. How could this be?

This is where I was impressed. When checking the status of the pool, a message status was shown that stated there were devices that experienced an unrecoverable error. AND there was given an action item for us to perform, basically to see if we needed to replace the device. A scrub was done to clean the errors. The demo also will take you through a quick replacement for the corrupted disk through resilvering.

This demo, while there is no audio is quick to go through and will get you acclimated to some of ZFS's features.

Go to Self Healing with ZFS to watch this demo.