One problem I noticed was after I'd replaced a failed drive (I chose a 3TB WD Red HDDs for replacements) I found that another drive had failed. Upon replacing this, again another drive failed. This left me with 3 x 3TB working drives and 1 x 2TB failed drive.
Now for the weird part...
If I remove and replace the 2TB drive with a new 3TB drive, the RAID set dies, the NAS won't boot successfully and all my data is missing. If I keep the 2TB in, the system seems to work fine once booted (I think). However, if I reboot it, all my custom app configurations disappear, for example Sick Beard, CouchPotato, NZBGet, Optware, etc, and it begins a RAID rebuild process that seems to get to 80% and then freeze the NAS. Weird, I know.
So anyway, the purpose of this post was to describe a hack I found which allowed me to 'buy more time' with my current '1 x failed drive' configuration so that the rebuild process doesn't get to 80% for a very long time, while I save some extra money to buy some spare 3TB drives to copy my data off onto.
The hack involved tricking the NAS to ignore the rebuild by changing the CPU timing of the RAID device to a very low number. This is done by following these steps:
1. Open /proc/sys/dev/raid/speed_limit_maxI found this allowed me to use the NAS for several days without interruption. I've since upgraded the firmware on the NAS and had to repeat the process again, however it appears to have still worked.
2. Change the default value (200000) to 5000
3. Save the file
4. Open /proc/sys/dev/raid/speed_limit_min
5. Change the default value (200000) to 5000
6. Save the file
7. Reboot the NAS
Now I just need to buy the replacement drives!
The article where I found out about this technique was here:
http://www.corelgott.de/?p=18
I hope this is useful as a temporary measure for some of you out there.
Cheers!