Disk spewing errors, performance tanking. Advice?

[sincero]

 

It looks like I should replace Disk 4 which has terrible performance and is barely able to read data from (though, I can.. I did manage to copy a few files off it to test). Disk 2 seems fine from a performance standpoint. unRAID has not marked Disk 4 as bad. However, the syslog has a ton of sector read errors. It has to be replaced, clearly.

 

How should I proceed? Should I manually remove this disk from the array and let parity take over? The disk is not marked as bad. Can I buy a couple 8TB disks to take over? This should give me a bit more room and get some of the ageing disks out of the array. Or should I shut it down and replace them then?

 

I have a full backup at home on another machine, so there is no concern there. But it would be nice to avoid needing to copy all 14TB back over gigabit slowly 

Edited September 29, 2023 by sincero

[sincero]

promathia-diagnostics-20230929-0159.zip

[JorgeB]

Disk4 has a lot of reallocated sectors, and the errors are logged as a disk problem, probably should just replace it now, but you can also run an extended SMART test to confirm, disk2 should also be replaced when possible, since it has a failing now SMART attribute, though one that usually doesn't cause read errors.

[itimpi]

It is also worth pointing out that disk1 and disk2 are still using reiserfs which is now deprecated.    Maybe now is the time to consider getting these converted to one of the other file system types supported by Unraid.

[sincero]

7 hours ago, JorgeB said:

Disk4 has a lot of reallocated sectors, and the errors are logged as a disk problem, probably should just replace it now, but you can also run an extended SMART test to confirm, disk2 should also be replaced when possible, since it has a failing now SMART attribute, though one that usually doesn't cause read errors.

 Thanks! So if I understand correctly, I can proceed with a parity swap once I order a new disk?

[sincero]

7 hours ago, itimpi said:

It is also worth pointing out that disk1 and disk2 are still using reiserfs which is now deprecated.    Maybe now is the time to consider getting these converted to one of the other file system types supported by Unraid.

 

Thanks! I'm planning maybe to upgrade over half the disks anyway (or maybe replace them all with 8TB) so I think this should take care of it since they'll have all new filesystems.

 

I'll probably just wait until Black Friday for some of them.

[JorgeB]

23 minutes ago, sincero said:

I can proceed with a parity swap once I order a new disk?

You can but that's only needed if the replacement disk is larger than current parity.

[sincero]

4 minutes ago, JorgeB said:

You can but that's only needed if the replacement disk is larger than current parity.

 

All the disks in this machine are from 2014/2015. I figure they're probably going to all be dying over the next few years so I'm considering whether it makes sense to replace them all with 6TBs or 8TBs to get fresh drives and another 5-10 years out of them. Thoughts?

 

My server lately grows at about 1.2TB per year, so I'll need more space eventually anyhow.

Edited September 29, 2023 by sincero

[JonathanM]

On 9/29/2023 at 11:47 AM, sincero said:

I think this should take care of it since they'll have all new filesystems.

Unclear if you are aware that rebuilt disks must have their original filesystems, as the filesystem is part of the emulation. To get a new filesystem you must add a new disk, or format an existing disk. Formatting erases all the data. There is a sticky thread discussing methods of moving data around to change filesystems.