I want to better understand hardware failure scenarios

[theothermatt_b]

It feels like answers to these questions are in a dozen different places, some older than others and may not apply. I don't want to be scared of hardware failure due to lack of knowledge, so I'd like to just ask about the different scenarios here.

 

Flash drive dies

I believe the solution here is to already have a backup of your flash drive, either

• through Unraid Connect (according to Unraid, this should backup any flash drive changes within a minute or two of them happening, so your backup should ideally be current),
• through manually backing it up, or
• through a backup plugin like Appdata Backup (which backs up your flash drive onto your array, although without a working flash drive i'm not sure how you get to the backup... hm.) '

What's the best solution here?

 

Array drive dies

Get a new drive of the same size or larger, and let the parity drive rebuild the data. I believe any data on the dead drive is completely inaccessible until you replace the drive. Is this accurate?

 

Parity drive dies

I believe you can get to all your data, but it's not "safe" until you get a new parity drive and rebuild it. Accurate?

 

Cache drive dies

Assuming you don't have two cache drives in a raid setup, you should be using a backup plugin like Appdata Backup to backup your appdata folder, otherwise you've just lost your appdata folder and with it all or most settings in any docker containers. The docker container settings that you can get to through Unraid are fine, but things like movie collections in plex, or indexer lists in jackett, are all potentially gone. (This is the one that happened to me, that made me want to make this thread) Is this correct?

 

Motherboard dies

If you can't (or don't want to) get the same motherboard... can you just plug all your hard drives and flash drive into a new motherboard, and everything will boot up just fine? That seems too simple. I feel like Unraid wouldn't get the drives in the right order or something? What's the solution here?

Edited December 7, 2023 by theothermatt_b

[itimpi]

1 hour ago, theothermatt_b said:

Array drive dies

Get a new drive of the same size or larger, and let the parity drive rebuild the data. I believe any data on the dead drive is completely inaccessible until you replace the drive. Is this accurate?

No.   Unraid will be emulating the drive and it will appear to be present throughout the rebuild and you will be able to read/write it as though the drive had not failed.   Performance will be degraded during the rebuild and you are not protected until the rebuild completes.

 

1 hour ago, theothermatt_b said:

Parity drive dies

I believe you can get to all your data, but it's not "safe" until you get a new parity drive and rebuild it. Accurate?

Correct (unless you have 2 parity drives).

 

1 hour ago, theothermatt_b said:

Cache drive dies

Assuming you don't have two cache drives in a raid setup, you should be using a backup plugin like Appdata Backup to backup your appdata folder, otherwise you've just lost your appdata folder and with it all or most settings in any docker containers. The docker container settings that you can get to through Unraid are fine, but things like movie collections in plex, or indexer lists in jackett, are all potentially gone. (This is the one that happened to me, that made me want to make this thread) Is this correct?

Yes.

 

1 hour ago, theothermatt_b said:

Motherboard dies

If you can't (or don't want to) get the same motherboard... can you just plug all your hard drives and flash drive into a new motherboard, and everything will boot up just fine? That seems too simple. I feel like Unraid wouldn't get the drives in the right order or something? What's the solution here?

This normally ‘just works’.  Unraid does not care how drives are connected as it recognises them by serial number.   I think this is one area where people are pleasantly surprised with how psinless it can be.
 

Things that can go wrong (but are normally easily worked around) is if one motherboard had a RAID disk controller and the other an HBA  so the disks were presented with different serial numbers of different sizes.  

 

VMs with passed through hardware might need to be handled carefully as the IDs of the passed through hardware will almost certainly change

[theothermatt_b]

Thank you for all of that!

 

So it sounds like the "you should have been doing this ahead of time" things are:

1) have your flash drive getting backed up routinely either by a plugin or through Unraid Connect

2) backup your appdata folder regularly through a plugin

 

Other than that, the system has fail-safes already in place for most other single-piece-of-hardware failures.

[itimpi]

Historically the appdata backup plugin could go both jobs.    More recently the Unraid Connect feature had given a web based option for the flash backup.