I've got a very basic Server 18.04 setup running that acts as my NAS/webserver/etc. Recently I picked up two HGST 1TB drives to add another RAID-1 array to the machine. The drives are 5400 RPM with 8 MB cache, so nothing spectacular, but they were cheap. I noticed when building the array, mdadm was resyching at just over 1MB/s and it was going to take something like three days to finish; obviously something was up. I suspected bad drives and began some testing to find out what was going on.

First some basic read tests with hdparm; hdparm -Tt /dev/sd<x> showed both drives reading at around 40MB/s. Not spectacular by any means. The much more interesting test was write speeds. Moving on to a write test using dd of=/dev/zero if=/dev/sd<x> bs=8k count=<y>, I saw interesting behavior. If I limited the count to under 1000 (meaning under 8MB of data written - the drive's cache size) I was getting great speeds, up to around 100MB/s. As soon as I went over 1000, though, write speeds plummeted to anywhere from 1MB/s, to 6MB/s if I was lucky.

So obviously at this point I'm thinking it's bad hard drives. Just to make sure it wasn't a drive bay issue, I swapped a known good drive (but still an older, slower one) into the same bay and got much more reasonable results - 80MB/s read from hdparm and 50MB/s from the same dd command.

Just because these were new-ish drives (manufacturer refurb units but still...) I wanted to double check my results to see if it wasn't something else being weird. I hooked up one of them to my windows machine via a USB-SATA adapter, created a simple NTFS volume, and ran CrystalDiskMark with a 500 MB test. It showed perfectly fine speeds for a USB2 SATA adapter - 100 MB/s read and 50 MB/s write. Certainly way better than the 1MB/s I saw on Ubuntu. The other new drive was also just as fine. Just to triple-check everything, I tried the control drive from earlier and it actually performed worse on Windows - only 25 MB/s read & write.

So here's the question: what could cause a drive to have poor performance on my Ubuntu system but be fine (as far as I can tell) on a Windows system? The Ubuntu machine is a Supermicro X8SIL motherboard (bios v1.1) with a Xeon x3470 and an on board RAID controller. No other drives hooked up to the system show this issue; just the two new ones.

I'm still a relative novice when it comes to advanced problems like this so any & all help is appreciated!

  • Do you have the current BIOS version R 1.2a ? Please click edit to add that fact to your question. Please do not click Add Comment, please use edit instead. Editing tips are at unix.stackexchange.com/editing-help#code .
    – K7AAY
    Aug 15 '19 at 23:55
  • 1
    I'm going to add a comment as well, because it's additional information that's specifically pertinent to your response :) I attempted an update to the latest bios (1.2) multiple times but it never properly booted. I was unable to access the bios menu or boot into the OS drive with 1.2. The only way to get anything other than a black screen was to remove all boot media except for a DOS USB drive and reflash to BIOS v1.1. Once back on 1.1, the system was once again able to boot and I could access the BIOS settings just fine.
    – Patrick
    Aug 17 '19 at 23:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.