You probably won't know that a drive has failed until you reboot the server. The drives will tend to function until this occurs so it's not a bad idea to reboot every so often just to check the integrity of the drives. Upon rebooting the server, the array will fail to start as if a drive is missing. The web GUI will indicate which drive is missing. The drive is identified by the brand and serial number. Shut down the array and locate the failed drive. I put small labels on the front of each drive tray to identify which drive is which so it's easy to locate them later.
Remove the faulty drive and swap it with one that has already been pre-cleared. I like to keep a spare drive on hand and pre-clear it ahead of time in case I need to replace one. Just connect the drive to a spare SATA port if you have one available and run pre-clear in the background, either using the command line prompt directly on the server using a keyboard and monitor or using the plug-in and run it from the web GUI. Pre-clearing not only prepares the drive for use in the array but also tests it to make sure it is fully functional. Search for pre-clear on the unRAID forum and you'll find out how to run it.
You can actually install a drive in unRAID without having to pre-clear it if you're replacing a failed drive as every location on the new drive will be overwritten anyway. If you're adding a drive to an existing array it must be pre-cleared first. Install the new drive in the slot and fire up the array. Go to the web GUI and assign the new drive to the slot previously occupied by the bad drive. Start the array and it will automatically rebuild the data from parity. You'll be presented with a prompt or two along the way so just answer yes to all of them and let the array do it's thing. Note that you can still access data on the array while data is being rebuilt. You just can't add any more data until the data rebuild cycle is complete.
Originally Posted by AVTechMan
The server does have IPMI so what options are available to flash the drive bay light?
Yep I have another parity drive ready to go. What I plan to do since all the drive bays are occupied I will probably remove the drive that's erroring out and put in the new parity drive.
You can certainly do this, but you'll have to delete the configuration file on the flash drive and start with a new configuration. I believe you can also indicate that it's a new configuration in the web GUI. You can assign the drives to whatever slot you want since it's going to rebuild parity from scratch anyway. The latest version of unRAID supports up to 30 drives and two parity drives. If you have an empty PCI slot you can install an simple 2-port SATA controller and add a new parity drive that way. You can also use an SATA port on the motherboard if you have one available. I got rid of the stock motherboard and other hardware as soon as I got my Supermicro server and replaced it with my own hardware. I use a standard PC PSU instead of the power supply that screams like a banshee that came with it.
You will have to rebuild the data using the existing configuration before you can update the parity drive and add a second one. If your new parity drive is the same size as the rest of your drives I would recommend keeping the new drive as a spare and getting a larger drive to use as the parity drive. This will allow for much easier expansion or upgrades down the road. You'd have to upgrade both parity drives if they're the same size as the rest of the drives in the array before you could upgrade to a larger drive elsewhere in the array. I'm not sure how large you can go with the drives using the stock SATA controllers. That's one of the reasons I swapped my setup out with my own hardware. My SATA controllers will support drives as large as 4TB and probably larger. I'm up to 30 drives (28 data, one parity, and one cache) for a current capacity of about 89.5TB.