July 2, 201115 yr Hi, newbie to unraid here. I have intermediate skills in linux sys administration. Just installed unraid 4.7 last night on a hp microserver with a 2tb western digital green drive 5200rpm. No parity drive as yet. Hdparam reported an average of 120mb/s transfer rate which is inline with posts ive found on these forums and other sources on the net. Plugged in another 1tb western digital green drive 5200rpm running on ntfs. Mounted as per the unoffical wiki recommendations and evereything works as expected. Hdparam also reported an average of 80mb/s transfer rate. df output - buffered reads on the 1tb green target read drive: root@files:/mnt# hdparm -t /dev/sdd1 /dev/sdd1: Timing buffered disk reads: 216 MB in 3.01 seconds = 71.87 MB/sec root@files:/mnt# hdparm -t /dev/sdd1 /dev/sdd1: Timing buffered disk reads: 254 MB in 3.01 seconds = 84.42 MB/sec root@files:/mnt# hdparm -t /dev/sdd1 /dev/sdd1: Timing buffered disk reads: 254 MB in 3.01 seconds = 84.48 MB/sec Started using cp and. Midnight commander to copy files from the 1tb to the 2tb and mc reports 8mb/s on average. I have plans to move terabytes to my unraid tower but this is taking way too long! My eyes are bleeding. I have checked my bios settings and it is definately on ahci mode and sata 300 mode. No other network activity is happening on the box and no other file io to and from both drives. This is idealized writes to the target 2tb drive (worrying): root@files:/mnt# dd if=/dev/zero of=/mnt/disk2/testfile bs=1024 count=1000000 1000000+0 records in 1000000+0 records out 1024000000 bytes (1.0 GB) copied, 82.0981 s, 12.5 MB/s Can anyone give me some pointers? Or help? Thanks for your time. syslog.txt smart-1tb.txt smart-2tb.txt
July 2, 201115 yr Maybe lots of little files? How is the NTFS drive connected physically? Could you test if it's the reading or writing that's slow?
July 2, 201115 yr Author There are moderate proportions of smaller files but the 8mb/s reading was over a very large dvd backup rip over 4 gigs large. The ntfs based 1tb green drive is connected via the onboard sata. So is the other drive. Could you clarify your last request? How can i test that? I did copy a file from md1 (riserfs) to md2 (risherfs) and it was quite fast! Will post speed in a sec. It was definately not 8mb/s!
July 2, 201115 yr Something like... idealized writes: dd if=/dev/zero of=/mnt/diskX/testfile bs=1024 count=1000000 unbuffered reads: hdparm -t /dev/mdX This is just for "how fast will it go". There's also "what's it doing?" How's memory and the processor doing when you copy? I don't know mc well. One DVD of files shouldn't do it but if you're copying huge numbers of files it might be dominating resources. Have you tried a subset of the copy, or a plain cp for comparison?
July 2, 201115 yr Author Something like... idealized writes: dd if=/dev/zero of=/mnt/diskX/testfile bs=1024 count=1000000 unbuffered reads: hdparm -t /dev/mdX This is just for "how fast will it go". There's also "what's it doing?" How's memory and the processor doing when you copy? Hi cyrnel, i have updated the original post. Load averages are high about 3, but CPU util is low, it's mainly waiting and idling. Memory is all used up but that's just linux buffering disk access?
July 2, 201115 yr Still no parity drive involved? You're sure nothing else is going on with that drive? Can we see what SMART thinks about it? http://lime-technology.com/wiki/index.php?title=Troubleshooting#Running_a_SMART_test
July 2, 201115 yr Author Still no parity drive involved? You're sure nothing else is going on with that drive? Can we see what SMART thinks about it? http://lime-technology.com/wiki/index.php?title=Troubleshooting#Running_a_SMART_test Most definitely nothing happening while im running those tests or when i am copying. No parity drive configured! I have the smart reports attached. Thanks for your time. smart-2tb.txt smart-1tb.txt
July 2, 201115 yr BTW, I just looked at the end of your syslog. I'd listen to it and add the nls=utf8 option to your mount command. ...and thanks for posting the report. Did you run a test or just a query? With that I'm done for the evening. Plenty of (smarter/more awake) people here to help though.
July 2, 201115 yr Author Running the idealized writes is really alarming!! root@files:~# dd if=/dev/zero of=/mnt/disk2/testfile bs=1024 count=1000000 1000000+0 records in 1000000+0 records out 1024000000 bytes (1.0 GB) copied, 95.9978 s, 10.7 MB/s root@files:~# dd if=/dev/zero of=/mnt/disk2/testfile bs=1024 count=1000000 1000000+0 records in 1000000+0 records out 1024000000 bytes (1.0 GB) copied, 81.5295 s, 12.6 MB/s root@files:~# df Filesystem 1K-blocks Used Available Use% Mounted on /dev/sdc1 3903488 56260 3847228 2% /boot /dev/md1 1953454928 398843504 1554611424 21% /mnt/disk1 /dev/md2 1953454928 184004472 1769450456 10% /mnt/disk2 While running dd this is the output of top top - 19:23:19 up 11 min, 2 users, load average: 0.47, 0.47, 0.36 Tasks: 74 total, 1 running, 73 sleeping, 0 stopped, 0 zombie Cpu(s): 0.1%us, 1.8%sy, 0.0%ni, 75.5%id, 22.5%wa, 0.0%hi, 0.2%si, 0.0%st Mem: 1001648k total, 361700k used, 639948k free, 10892k buffers Swap: 0k total, 0k used, 0k free, 307436k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 116 root 20 0 0 0 0 D 2 0.0 0:02.76 sync_supers 1448 root 20 0 0 0 0 D 2 0.0 0:00.05 flush-9:2 1340 root 20 0 0 0 0 D 0 0.0 0:00.00 reiserfs/0 1450 root 20 0 2108 880 700 R 0 0.1 0:00.00 top Something is very wrong! Help!
July 2, 201115 yr Author Just read the sticky post regarding western digital EARS drives at http://lime-technology.com/forum/index.php?topic=11842.0. My 2tb is a EARX but what the hey, i'm going to preclear it and set it to mbr-4k and see what happens. It's running in the background as i type. 1% 39gig out of 2000gig, and only on step one, and it's blazing at 120MB/s - oh dear.. this might take a while...
July 2, 201115 yr Author Pre-clear performed on the 2tb western digital EACX last night. Took a very long time. The command i used was preclear_dish.sh -A -n /dev/sdb dd has improved from 8-10MB/s to 30MB/s. Still not up there at the 100MB/s level.. root@files:~# dd if=/dev/zero of=/mnt/disk2/testfile bs=1024 count=1000000 1000000+0 records in 1000000+0 records out 1024000000 bytes (1.0 GB) copied, 34.5666 s, 29.6 MB/s root@files:~# dd if=/dev/zero of=/mnt/disk2/testfile bs=1024 count=1000000 1000000+0 records in 1000000+0 records out 1024000000 bytes (1.0 GB) copied, 30.2665 s, 33.8 MB/s This is what fdisk on it looks like now: root@files:~# fdisk /dev/sdb The number of cylinders for this disk is set to 243201. There is nothing wrong with that, but this is larger than 1024, and could in certain setups cause problems with: 1) software that runs at boot time (e.g., old versions of LILO) 2) booting and partitioning software from other OSs (e.g., DOS FDISK, OS/2 FDISK) Command (m for help): p Disk /dev/sdb: 2000.3 GB, 2000398934016 bytes 255 heads, 63 sectors/track, 243201 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System /dev/sdb1 1 243202 1953514552 0 Empty Partition 1 does not end on cylinder boundary. Command (m for help): ^C This is what fdisk used to look on it. (this is an identical brand new WD 2tb EACX drive which has not been precleared) root@files:~# fdisk /dev/sda The number of cylinders for this disk is set to 62016336. There is nothing wrong with that, but this is larger than 1024, and could in certain setups cause problems with: 1) software that runs at boot time (e.g., old versions of LILO) 2) booting and partitioning software from other OSs (e.g., DOS FDISK, OS/2 FDISK) Command (m for help): p Disk /dev/sda: 2000.3 GB, 2000398934016 bytes 1 heads, 63 sectors/track, 62016336 cylinders Units = cylinders of 63 * 512 = 32256 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System /dev/sda1 2 62016336 1953514552+ 83 Linux Partition 1 does not end on cylinder boundary. Command (m for help): ^C Now copying a file from one riserfs based drive and the newly precleared EACX 2tb is averaging 25MB/s measured using midnight commander on a 5gig mkv file.
July 2, 201115 yr Author BTW, I just looked at the end of your syslog. I'd listen to it and add the nls=utf8 option to your mount command. ...and thanks for posting the report. Did you run a test or just a query? With that I'm done for the evening. Plenty of (smarter/more awake) people here to help though. Those slow speeds i quoted previously, were performed while the system was doing nothing. Both running the tests and using cp. Help! *gulp*
July 3, 201115 yr Any improvement today? Evidently those EARX drives are just EARS drives with a 6Gb/s interface, so yes, preclear with the -A option if your default isn't already set. I have two WD20EARS drives in my main server. The same poor-man's test on several of my drives: (hdparm -t for reads, 1GB dd>file for writes) (System has a parity drive so write numbers are affected.) WD20EARS1 WD20EARS2 SamsungHD204UI Hitachi HDT72101 Hitachi 7K300 What 2TB 5400 2TB 5400 2TB 5400 1TB 7200 2TB 7200 Card PCIe1a PCIe1b PCIe2b OnBoard OnBoard Full 95% 55% 98% 80% (Parity) Write Read Write Read Write Read Write Read Write Read Test1 31.8 103.59 39.1 103.79 45.6 145.66 42.2 100.71 NA 163.84 Test2 30.3 103.63 42.9 105.22 43.1 146.34 41.0 101.02 NA 164.12 Test3 31.9 104.93 39.5 105.04 47.4 146.09 44.4 100.70 NA 163.97 Test4 32.4 103.50 43.7 105.16 43.4 146.17 41.1 100.84 NA 164.05 Test5 31.3 103.82 43.1 105.83 46.9 146.24 44.4 100.91 NA 163.90 If the relative speeds of my EARS drives are due to track position (fullness) then your new drive should be faster. Could you try swapping interfaces with another drive to see if the slowness follows?
Archived
This topic is now archived and is closed to further replies.