Slow Parity Rebuild after installing new drive

[PhilUK]

Hi all,

 

I'm running a Gen8 microserver - 5 drives in the array in total. 4 x 3TB running in the main drive bays and one 240GB SSD (the aim is to retire this shortly and  move the data to a protected cache) running through a BR10i flashed to LSI firmware.

All been running great but I have nearly run out of space so need to replace 2 of the 3TB disks with 2x16TB drives.

 

I have precleared 1x disk to test for any issues and it has completed successfully. I have therefore stopped the array, and swapped out the current 3TB drive for the 16TB drive.

 

This has triggered a parity rebuild which is going VERY slow - started at 6MB/s, now at 5MB/s with 32 days remaining...

 

Anything I can do to diagnose the issue as I believe this should be running much faster than this? - I've done a search and looked in the logs but can't see anything that stands out at the moment.

 

I'll post my diagnostics and any pointers would be very much welcomed.

 

Thanks

[PhilUK]

tower-diagnostics-20240908-1317.zip

[JorgeB]

Nothing obvious logged, make write cache is enabled in the server BIOS, if that doesn't help run the diskspeed docker to confirm all disks are performing normally.

[PhilUK]

Thanks Jorge, diskspeed has sent me down the right path!

 

Looks like I may have an issue with the 240GB SSD for some reason - not giving any errors at the moment but...:

I'm copying the files on this drive off onto a spare USB drive I have using Midnight Commander via the terminal

It's only running at 5-6MB/s. Diskspeed shows speeds on all the other spinning disks are running ok.

 

I've tried changing the cable it's using on the LSI card - no dice

I've tried running the drive from the onboard SATA port on the motherboard - no dice

I've tried changing the SATA power cable - no dice.

 

At this stage, I'm thinking to simply move the data (slowly) from the SSD into the other drives on the array then do a new config excluding the SSD from the array - would there be any issues with this (apart from the fact I currently have no parity protection).

 

Thanks again

[PhilUK]

OK, so I have moved all files off the 240GB SSD which was part of the array into the cache pool, reset the configuration and removed the SSD from the array.

 

Parity rebuild has started - running at 100+ MB/s - problem solved! Not sure what the issue with the SSD is - maybe age and lack of TRIM support on the controller but glad it's sorted now.

 

Now to allow the parity check to complete and preclear the next 16TB drive!