What is your parity check speed?

Mine starts at about 140mb/s and ends on about 74mb/s. I have mostly 3TB WD Reds but two 4TB WD Reds. Looking forward to seeing what its like after I replace them all with the 6TB WD Reds over the next month.

Mine starts at about 140mb/s and ends on about 74mb/s. I have mostly 3TB WD Reds but two 4TB WD Reds. Looking forward to seeing what its like after I replace them all with the 6TB WD Reds over the next month.

That will indeed be interesting.    With a 20% higher sustained data rate; and only a single "inner cylinder slowdown" to contend with [right now your check slows down as the 3TB drives approach their inner cylinders; then again when the 4TB units do],  I'd expect a very nice improvement in the average speed.  [Total time will, of course, increase, since you'll have 2TB more space to check]

5 seagate ST3000DM001 3TBs, 4 array, 1 parity.

Last checked on Mon Sep 1 10:58:02 2014 BST (today), finding 0 errors.

? Duration: 5 hours, 58 minutes. Average speed: 139.7 MB/sec

I recently removed the last 2TB drive from my array. So, I'm running all HGST/Hitachi 4TB 7200rpm drives.  Twelve drives in total, 11 + parity. 

My parity check completed this morning at 117.96MB/s.

I'm pretty consistent with the following.

Machines are:

unraid - N40L Running bare metal 4GB unRAID 5

unraid1/unraid2 N54L running under ESX 5 4GB unRAID 5

Drive models are:

4TB are Seagates 54(9?)00

3TB are Seagate 7200 RPM and/or Hitachi 5400 RPMs

2TB are all WD Green drives.

root@unRAID:/mnt/disk3# grep -i sync /var/log/syslog | grep done

Mar 27 08:48:41 unRAID kernel: md: sync done. time=31719sec

Apr 27 08:38:51 unRAID kernel: md: sync done. time=31129sec

May 27 08:37:55 unRAID kernel: md: sync done. time=31074sec

Jun 27 08:50:30 unRAID kernel: md: sync done. time=31829sec

Jul 27 08:38:54 unRAID kernel: md: sync done. time=31133sec

Aug 27 08:41:01 unRAID kernel: md: sync done. time=31260sec

root@unRAID:/mnt/disk3# df -vH /mnt/disk*

Filesystem            Size  Used  Avail Use% Mounted on

/dev/md1              3.1T  2.6T  492G  84% /mnt/disk1

/dev/md2              3.1T  3.0T    88G  98% /mnt/disk2

/dev/md3              3.1T  3.0T    28G 100% /mnt/disk3

root@unRAID1:~# grep -i sync /var/log/syslog | grep done

Sep 19 03:03:16 unRAID1 kernel: md: sync done. time=31171sec

Feb 27 08:49:06 unRAID1 kernel: md: sync done. time=31744sec

Feb 28 07:10:14 unRAID1 kernel: md: sync done. time=31947sec

Mar 27 08:32:07 unRAID1 kernel: md: sync done. time=30726sec

Apr 27 08:34:18 unRAID1 kernel: md: sync done. time=30857sec

May 27 08:41:06 unRAID1 kernel: md: sync done. time=31264sec

Jun 27 08:49:16 unRAID1 kernel: md: sync done. time=31755sec

Jul 27 08:45:57 unRAID1 kernel: md: sync done. time=31555sec

Aug 27 08:44:32 unRAID1 kernel: md: sync done. time=31471sec

root@unRAID1:~# df -vH /mnt/disk*

Filesystem            Size  Used  Avail Use% Mounted on

/dev/md1              2.1T    34M  2.1T  1% /mnt/disk1

/dev/md2              2.1T    34M  2.1T  1% /mnt/disk2

/dev/md3              2.1T  1.3T  759G  63% /mnt/disk3

root@unRAID2:~# grep -i sync /var/log/syslog | grep done

Feb 27 08:18:17 unRAID2 kernel: md: sync done. time=29895sec

Feb 27 16:36:28 unRAID2 kernel: md: sync done. time=29890sec

Mar 27 08:18:12 unRAID2 kernel: md: sync done. time=29890sec

Apr 27 08:18:30 unRAID2 kernel: md: sync done. time=29908sec

May 27 08:18:12 unRAID2 kernel: md: sync done. time=29890sec

Jun 27 08:18:03 unRAID2 kernel: md: sync done. time=29881sec

Jul 27 08:18:15 unRAID2 kernel: md: sync done. time=29893sec

Aug 27 08:17:49 unRAID2 kernel: md: sync done. time=29867sec

root@unRAID2:~# df -vH /mnt/disk*

Filesystem            Size  Used  Avail Use% Mounted on

/dev/md1              4.1T    34M  4.1T  1% /mnt/disk1

/dev/md2              4.1T  1.5T  2.6T  36% /mnt/disk2

/dev/md3              4.1T  3.1T  958G  77% /mnt/disk3

Last Parity check on 08/01/2014 using 1,536k window, over a total of 2,930,266,532 blocks

Duration:  10:03:02 (36,182 secs).  Avg Speed: 79.09 MB/sec.

 

Specs in sig.

Finally got around to using the unraid tunables tester script on my server.  4MB/sec speed increase:

Parity check on 09/03/2014 08:22:51 using 3,584k window, over a total of 2,930,266,532 blocks

Duration:  09:34:00 (34,440 secs).  Avg Speed: 83.09 MB/sec.

Parity 6TB WD Red, 1 x 4TB WD Red Data, 4 x 3TB WD Red Data. Duration: 16 hours, 18 minutes, 1 second. Average speed: 102.3 MB/sec.

Now comes the fun part changing all my 3TB and the 4TB disks for 6TB's. Im hoping to see a speed increase once all the disks are replaced with 6TB WD Red's.

Average is usually around 137 MB/s.

What I've seen in disc speeds is that you can expect near 100% of the speed for the 1st 1/3rd of the drive and minor slowing for the 2nd 1/3rd, and a rapid drop for the last 1/3rd.

When 64bit UNRAID was released, I ran a benchmark comparing a 32bit parity check vs a 64bit parity check. The 64bit version ran slightly faster.

http://strangejourney.net/UNRAID/32vs64.html

System has a mix of 3TB and 4TB Seagate drives.

Your parity check will be only as fast as your slowest drive. If you're looking to increase your parity check speeds, use a tool I wrote to speed test each of your drives.

http://lime-technology.com/forum/index.php?topic=31073.0

Parity 6TB WD Red, 1 x 4TB WD Red Data, 4 x 3TB WD Red Data. Duration: 16 hours, 18 minutes, 1 second. Average speed: 102.3 MB/sec.

Now comes the fun part changing all my 3TB and the 4TB disks for 6TB's. Im hoping to see a speed increase once all the disks are replaced with 6TB WD Red's.

Yes, you should see a VERY nice increase when you switch to all 6TB drives.    You're currently slowing down significantly for the inner tracks of the 3TB drives;  then the inner tracks of the 4TB drive; and finally for the inner tracks of the 6TB drive.    Once you eliminate the 3 & 4 TB drives, you'll get superb speeds until about the last TB or so of the 6TB units.

My Athlon X2-250u with Seagate NAS and desktop 4TB drives gets 90 or so MB/s, my ASUS C60 with cheap old drives gets 30-40 

Last run

 
Last checked on Monday, 15 September 2014, 13:28 (27 days ago), finding 0 errors. 
* Duration: 11 hours, 28 minutes, 3 seconds. Average speed: 72.7 MB/sec

Specs are in signature

Direct copy from Dashboard of Dynamix:

Last checked on Sun Sep 7 12:47:14 2014 BST (35 days ago), finding 0 errors.

Duration: 5 hours, 27 minutes, 10 seconds. Average speed: 152,9 MB/sec

Parity 6TB WD Red, 1 x 4TB WD Red Data, 4 x 3TB WD Red Data. Duration: 16 hours, 18 minutes, 1 second. Average speed: 102.3 MB/sec.

Now comes the fun part changing all my 3TB and the 4TB disks for 6TB's. Im hoping to see a speed increase once all the disks are replaced with 6TB WD Red's.

 

Yes, you should see a VERY nice increase when you switch to all 6TB drives.    You're currently slowing down significantly for the inner tracks of the 3TB drives;  then the inner tracks of the 4TB drive; and finally for the inner tracks of the 6TB drive.    Once you eliminate the 3 & 4 TB drives, you'll get superb speeds until about the last TB or so of the 6TB units.

So after removing the WD 4TB Reds I'm left with 3 x 6TB and 2 x 3TB WD Reds. The latest parity check was Duration: 15 hours, 1 minute, 6 seconds. Average speed: 111.0 MB/sec.

So the time is now 1hr 17 minutes less and the average speed increased by 9 MB/sec. Just two more WD 6TB Red's to preclear and then the WD 3TB Red's can go.

Thanks for the update.    You basically gained an hour and 17 minutes just by eliminating the last 1TB worth of checking on the 4TB drive.    You should gain even more when you get rid of the 3TBs, since you'll not only gain for not having to process the inner cylinders of those drives (perhaps the last TB); but you'll also eliminate the throttling you now have while the 3TB are in the system because of their lower areal density.

Be sure to post the final results when you're running an "all 6TB" array 

Yes I will do that Gary. So far Im impressed with the changes the 6TB drives have made. It also means I can look at changing my 20 bay rack mount for a smaller and maybe a little bit quieter system. while keeping all my hardware except for the LSI 9201-16i HBA, I can change this to my LSI 9211-8i card.