February 13, 201412 yr I'm running 6.0-beta3. I'm not sure if what I'm seeing is normal and need some feedback. I'm moving a chunk of data between two 7200 RPM drives. Parity drive is also 7200 RPM. Note however that my parity is currently invalid (although unraid UI says it's valid for some reason). # mv /mnt/disk6/dir_of_large_files /mnt/disk2 # date; df -h /mnt/disk2 Thu Feb 13 11:22:03 EST 2014 Filesystem Size Used Avail Use% Mounted on /dev/md2 1.9T 515G 1.4T 28% /mnt/disk2 # date; df -h /mnt/disk2 Thu Feb 13 11:47:52 EST 2014 Filesystem Size Used Avail Use% Mounted on /dev/md2 1.9T 523G 1.4T 29% /mnt/disk2 So 8 GB in 25 minutes. That's 5.28 MB/s, which seems really bad. So I stopped the array, unassigned the parity drive, and tried again. # date; df -h /mnt/disk2 Thu Feb 13 11:53:09 EST 2014 Filesystem Size Used Avail Use% Mounted on /dev/md2 1.9T 524G 1.4T 29% /mnt/disk2 # date; df -h /mnt/disk2 Thu Feb 13 12:03:26 EST 2014 Filesystem Size Used Avail Use% Mounted on /dev/md2 1.9T 550G 1.3T 30% /mnt/disk2 26 GB in 10 minutes is 43 MB/s. That's more like what I was expecting. Does this seem like a 6.0 bug? Or is it normal?
February 13, 201412 yr Why do you say your parity is invalid? (I believe you, but what's going on to put you in this spot?)
February 13, 201412 yr Author Syslog attached. (Thanks!) I've been replacing drives, and at some point I did "New Config". At that time IIRC unraid started doing a parity check, but I interrupted it because I had more disk swaps to do. I figured when everything was done that I'd do a parity check. FWIW I did kick of a parity check earlier, and saw in the log that there were some corrections. I stopped the parity check during the times above, however. Edit: Really attaching syslog.txt. syslog.txt
February 13, 201412 yr Author Dunno if it helps, but after reinstating the parity drive, the parity check is going at >65 MB/s, with the CPU cores pegged by a HandBrake encode.
February 13, 201412 yr Author Test running bare-metal. Can you clarify? Do you mean by stopping unraid and manually mounting the disks? I can certainly do that, but I'm not sure what that will show, since I already got good speeds under unraid but without the parity drive.
February 13, 201412 yr Test running bare-metal. Can you clarify? Do you mean by stopping unraid and manually mounting the disks? He means test it by booting into unRAID without Xen. Per your syslog you are obviously running unRAID6 w/ Xen.
February 20, 201412 yr Author Just a quick follow-up. Before trying without Xen I wanted to finish a parity check and get a baseline for read and write performance. I couldn't reproduce the original problem. Read test: for n in d e g h f l k j;do hdparm -t /dev/sd$n;done - parity: 109.43, 106.42, 107.59 - disk 1: 132.46, 132.33, 132.59 - disk 2: 189.62, 189.62, 188.70 - disk 3: 82.72, 82.53, 82.59 - disk 4: 84.07, 84.11, 83.92 - disk 5: 104.22, 97.42, 116.62 - disk 6: 184.34, 185.43, 184.50 - disk 7: 118.82, 118.95, 118.64 Write test: for n in 1 2 3 4 5 6 7;do dd if=/dev/zero of=/mnt/disk$n/test.dd bs=1M count=256 conv=fdatasync;done - disk 1: 28.5, 33.4, 28.0 - disk 2: 32.7, 37.5, 33.5 - disk 3: 28.9, 32.7, 31.1 - disk 4: 7.9, 22.9, 25.5 - disk 5: 20.8, 28.1, 17.9 - disk 6: 28.3, 35.9, 29.9 - disk 7: 13.5, 13.8, 13.3 Disk 4 is a little weird with the write speeds, but I was moving the files from disk 5 to disk 2, so I don't think that's relevant.
Archived
This topic is now archived and is closed to further replies.