The mdadm in F-11 includes a new feature: automatic data scrubbing via a cron.weekly script. The original version of the raid-check cron job script told the raid subsystem to do a repair action instead of a check action. In the event of a check action, it will attempt to read all sectors on a device, and on read error it will attempt to reconstruct the errored read from redundant data. On a repair action it does the same thing, but in the event data blocks and parity blocks (for raid4/5/6) or in the event two data blocks (for raid1/10) don't match, it attempts to automatically pick the correct data block and synchronize the array. We don't want this. We only want to repair read errors. In the event of a data mismatch, we want to allow the user to manually inspect the two different versions of the data and do the right thing. So, this mdadm update only changes the raid-check cron job to perform a check instead of a repair operation.

Version of package: mdadm-3.0-0.devel3.7.fc11