Apacer AS350X 1TB SATA SSD, tested using CrystalDiskMark shows the expected read/write speeds for a SATA SSD (r - 500+MB/s / w - 400+MB/s).

it can sustain 300+MB/s when writing large files (4GB) to the SSD.

but in UNRAID (formatted BTRFS) it only gets about 40MB/s write speed.

Tested using dd

dd if=/dev/zero of=test1.img bs=1G count=10 oflag=dsync

Read speed is okay.

when it was part of a RAID1 pool, the pool's write was about 20MB/s and CPU cores go red when writing to the pool.

i thought the disk was faulty, but why does it seem fine in Windows?

i'm on Unraid 7.1.2

9 minutes ago, limawaken said:

it can sustain 300+MB/s when writing large files (4GB) to the SSD.

I doubt it will be able to sustain those speeds for long, but do a test with pv:

pv /path/to/large/file > /path/to/dest

And post the results, including the command used.

hi @JorgeB ,

ok, this is really embarrassing... i put the disk back and mounted using unassigned devices.

formatted to BTRFS again (because earlier i formatted it to NTFS to test in Windows). guess what... now the disk seems to be ok.

Here's the output of PV

root@SILOmetalico:~# pv /mnt/cache/isos/VeeamBackup\&Replication_11.0.0.837_20210525.iso > /mnt/disks/AD20230608A0101901/test.iso

6.72GiB 0:00:14 [ 477MiB/s] [==================================================================================================>] 100%

i also ran DD again several times for sanity

root@SILOmetalico:~# dd if=/dev/zero of=/mnt/disks/AD20230608A0101901/test1.img bs=1G count=10 oflag=dsync

10+0 records in

10+0 records out

10737418240 bytes (11 GB, 10 GiB) copied, 26.9105 s, 399 MB/s

root@SILOmetalico:~# dd if=/dev/zero of=/mnt/disks/AD20230608A0101901/test1.img bs=1G count=10 oflag=dsync

10+0 records in

10+0 records out

10737418240 bytes (11 GB, 10 GiB) copied, 28.2348 s, 380 MB/s

root@SILOmetalico:~# dd if=/dev/zero of=/mnt/disks/AD20230608A0101901/test1.img bs=1G count=10 oflag=dsync

10+0 records in

10+0 records out

10737418240 bytes (11 GB, 10 GiB) copied, 27.2 s, 395 MB/s

i'm really puzzled. it was definitely causing issues with my raid1 btrfs pool, so removed it, cleared it and reformatted using unassigned devices, and i swear it could only manage 40MB/s writes.

it seems like reformatting the disk in Windows (NTFS) had somehow fixed it.

really sorry for wasting your time, but i'm so curious now - what could have happened?

i actually had this problem before and threw out another SSD because i though it gone bad... i'm really regretting it now because at that time i didn't think of testing it in Windows before throwing it away

Edited June 18, 20251 yr by limawaken

1 hour ago, limawaken said:

really sorry for wasting your time, but i'm so curious now - what could have happened?

Not sure, was the other mirror the same model?

it was a 3 disk raid1 - 2 Apacer AS350X 1TB SSDs and 1 Sandisk Ultra 1TB SSD.

i was so certain the problem was with the disk because i removed it from the pool and tested it by itself (mounted and formatted using unassigned devices), confirming that it really did have write speed problems at that time.

i have replaced the "faulty" AS350x with a mx500, and write speed has improved however the pool only seems to be able to write at about 80MB/s.

i suspect the other AS350x is also having write speed problems because i have not seen it being written above 100MB/s, while i do see write speeds above 350MB/s for the other 2 disks.

i plan to swap it out when the new disk i ordered arrives.