Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have a Software RAID1 disk that is randomly experiencing very slow read times.

hdparm -t /dev/md0
Timing buffered disk reads: 2 MB in 12.43 seconds = 164.80 kB/sec

It appears to be one disk in the RAID set that is consistently slower than the other. They are both identical 1.5TB SATA drives though. I've read some tuning advice using hdparm, but it did not seem to apply to SATA drives and I was hesitant to experiment on a production system.

Any advice would be appreciated.

share|improve this question
What is the hdparm -tT of the constituent devices? – Mark Wagner May 12 '11 at 22:16
Why do you think one drive is slower? What test(s) have you run to identify that? – Slartibartfast May 13 '11 at 2:29
When I test the RAID array (hdparm -t /dev/md0), I see slow performance. When I test each individual drive, sda is fast (50Mbps), sdb is consistently slow (1Mbps or slower) – John P May 13 '11 at 14:33
Since it is a mirror, have you considered replacing sdb and re-testing? – Slartibartfast May 16 '11 at 4:47

It's quite possible that one drive is experiencing either errors or abnormal conditions such as high temperature.

Check the SMART logs for more information.

The tuning options in hdparm should apply to SATA drives as well, BTW.

share|improve this answer
Nothing unusual came out of the SMART logs - parameters for both drives were within a few percentage points of being the same – John P May 13 '11 at 2:13

If you want to eliminate the drives as the source of issues with read performance, do some simple tests with dd.

dd if=/dev/sda of=/dev/null bs=32M count=32
dd if=/dev/sdb of=/dev/null bs=32M count=32
dd if=/dev/md0 of=/dev/null bs=32M count=32

(of course, you can use the hdparm tool too / instead, but dd / application level testing may be more indicative of real-world performance, as long as you can simulate your actual load)

Read tests are non-destructive, so this shouldn't affect your array.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.