slow and jittery uploads on omv4 - odroid-xu4

    This site uses cookies. By continuing to browse this site, you are agreeing to our Cookie Policy.

    • slow and jittery uploads on omv4 - odroid-xu4

      greetings

      i am using omv4 latest official build for odroid-xu4
      im facing jittery uploads on FTP and constant 25MB/s uploads on SMB
      downloads are perfect 110MB/s
      tried iperf and its fine
      hdd benchmark is all fine too

      on omv3 never faced such issue

      i'm attaching GIF files for FTP and SMB
      here's the log

      Brainfuck Source Code

      1. ### hdparm results ###
      2. hdparm -Tt /dev/sda
      3. /dev/sda:
      4. Timing cached reads: 1818 MB in 2.00 seconds = 909.49 MB/sec
      5. Timing buffered disk reads: 382 MB in 3.01 seconds = 126.95 MB/sec
      6. ### dd results ###
      7. dd if=/dev/zero of=tempfile bs=1M count=1024 conv=fdatasync,notrunc
      8. 1024+0 records in
      9. 1024+0 records out
      10. 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 42.8506 s, 25.1 MB/s
      11. echo 3 > /proc/sys/vm/drop_caches && dd if=tempfile of=/dev/null bs=1M count=1024
      12. 1024+0 records in
      13. 1024+0 records out
      14. 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 7.86649 s, 136 MB/s
      15. iperf
      16. # bidirectional test individually
      17. C:\>iperf -c omv -i 2 -r
      18. ------------------------------------------------------------
      19. Server listening on TCP port 5001
      20. TCP window size: 208 KByte (default)
      21. ------------------------------------------------------------
      22. ------------------------------------------------------------
      23. Client connecting to omv, TCP port 5001
      24. TCP window size: 208 KByte (default)
      25. ------------------------------------------------------------
      26. [ 4] local 192.168.0.10 port 59790 connected with 192.168.0.16 port 5001
      27. [ ID] Interval Transfer Bandwidth
      28. [ 4] 0.0- 2.0 sec 217 MBytes 910 Mbits/sec
      29. [ 4] 2.0- 4.0 sec 218 MBytes 914 Mbits/sec
      30. [ 4] 4.0- 6.0 sec 217 MBytes 910 Mbits/sec
      31. [ 4] 6.0- 8.0 sec 214 MBytes 895 Mbits/sec
      32. [ 4] 8.0-10.0 sec 219 MBytes 919 Mbits/sec
      33. [ 4] 0.0-10.0 sec 1.06 GBytes 910 Mbits/sec
      34. [ 4] local 192.168.0.10 port 5001 connected with 192.168.0.16 port 54592
      35. [ 4] 0.0- 2.0 sec 220 MBytes 924 Mbits/sec
      36. [ 4] 2.0- 4.0 sec 215 MBytes 902 Mbits/sec
      37. [ 4] 4.0- 6.0 sec 193 MBytes 811 Mbits/sec
      38. [ 4] 6.0- 8.0 sec 203 MBytes 850 Mbits/sec
      39. [ 4] 8.0-10.0 sec 219 MBytes 920 Mbits/sec
      40. [ 4] 0.0-10.0 sec 1.03 GBytes 881 Mbits/sec
      41. # bidirectional test simultaneously
      42. C:\>iperf -c omv -i 2 -d
      43. ------------------------------------------------------------
      44. Server listening on TCP port 5001
      45. TCP window size: 208 KByte (default)
      46. ------------------------------------------------------------
      47. ------------------------------------------------------------
      48. Client connecting to omv, TCP port 5001
      49. TCP window size: 208 KByte (default)
      50. ------------------------------------------------------------
      51. [ 4] local 192.168.0.10 port 59960 connected with 192.168.0.16 port 5001
      52. [ 5] local 192.168.0.10 port 5001 connected with 192.168.0.16 port 54600
      53. [ ID] Interval Transfer Bandwidth
      54. [ 4] 0.0- 2.0 sec 198 MBytes 829 Mbits/sec
      55. [ 5] 0.0- 2.0 sec 150 MBytes 627 Mbits/sec
      56. [ 4] 2.0- 4.0 sec 189 MBytes 792 Mbits/sec
      57. [ 5] 2.0- 4.0 sec 171 MBytes 715 Mbits/sec
      58. [ 4] 4.0- 6.0 sec 198 MBytes 830 Mbits/sec
      59. [ 5] 4.0- 6.0 sec 172 MBytes 723 Mbits/sec
      60. [ 4] 6.0- 8.0 sec 193 MBytes 808 Mbits/sec
      61. [ 5] 6.0- 8.0 sec 174 MBytes 731 Mbits/sec
      62. [ 4] 8.0-10.0 sec 192 MBytes 807 Mbits/sec
      63. [ 4] 0.0-10.0 sec 970 MBytes 813 Mbits/sec
      64. [ 5] 8.0-10.0 sec 171 MBytes 717 Mbits/sec
      65. [ 5] 0.0-10.0 sec 838 MBytes 703 Mbits/sec
      Display All
      Images
      • FTP.gif

        995.24 kB, 812×702, viewed 41 times
      • SMB.gif

        383.05 kB, 796×574, viewed 40 times
    • kamilmirza wrote:

      same test from my ubuntu VM

      Sorry, no idea what you're doing. To be able to help please provide output from 'sudo armbianmonitor -u', then check the mountpoint of your sda partition, do a 'cd /path/to/sda1' and provide output from

      Source Code

      1. iozone -e -I -a -s 1000M -r 128k -r 16384k -i 0 -i 1
    • tkaiser wrote:

      Sorry, no idea what you're doing. To be able to help please provide output from 'sudo armbianmonitor -u', then check the mountpoint of your sda partition, do a 'cd /path/to/sda1' and provide output from

      Source Code

      1. iozone -e -I -a -s 1000M -r 128k -r 16384k -i 0 -i 1
      here's the armbianmonitor -u log LINK
      and here's the iozone out
      strange it shows 19-23 MBs
      benchmark the same HDD with NTFS it shows above 110 MB/s on Windows

      Source Code

      1. root@odroid:/srv/dev-disk-by-label-nas# iozone -e -I -a -s 1000M -r 128k -r 16384k -i 0 -i 1 [2/142]
      2. Iozone: Performance Test of File I/O
      3. Version $Revision: 3.429 $
      4. Compiled for 32 bit mode.
      5. Build: linux
      6. Contributors:William Norcott, Don Capps, Isom Crawford, Kirby Collins
      7. Al Slater, Scott Rhine, Mike Wisner, Ken Goss
      8. Steve Landherr, Brad Smith, Mark Kelly, Dr. Alain CYR,
      9. Randy Dunlap, Mark Montague, Dan Million, Gavin Brebner,
      10. Jean-Marc Zucconi, Jeff Blomberg, Benny Halevy, Dave Boone,
      11. Erik Habbinga, Kris Strecker, Walter Wong, Joshua Root,
      12. Fabrice Bacchella, Zhenghua Xue, Qin Li, Darren Sawyer,
      13. Vangel Bojaxhi, Ben England, Vikentsi Lapa.
      14. Run began: Thu Jul 12 00:48:31 2018
      15. Include fsync in write timing
      16. O_DIRECT feature enabled
      17. Auto Mode
      18. File size set to 1024000 kB
      19. Record Size 128 kB
      20. Record Size 16384 kB
      21. Command line used: iozone -e -I -a -s 1000M -r 128k -r 16384k -i 0 -i 1
      22. Output is in kBytes/sec
      23. Time Resolution = 0.000001 seconds.
      24. Processor cache size set to 1024 kBytes.
      25. Processor cache line size set to 32 bytes.
      26. File stride size set to 17 * record size.
      27. random random bkwd record stride
      28. kB reclen write rewrite read reread read write read rewrite read fwrite frewrite fread freread
      29. 1024000 128 6772 5820 75151 87619
      30. 1024000 16384 23092 19297 103037 120202
      31. iozone test complete.
      Display All
    • kamilmirza wrote:

      Here's the armbianmonitor -u

      Source Code

      1. /dev/sda1 /sharedfolders/downloads btrfs rw,relatime,space_cache,subvolid=5,subvol=/root/downloads 0 0
      2. /dev/sda1 /sharedfolders/nas btrfs rw,relatime,space_cache,subvolid=5,subvol=/root/nas 0 0
      3. /dev/sda1 /sharedfolders/root btrfs rw,relatime,space_cache,subvolid=5,subvol=/root 0 0
      4. /dev/sda1 /srv/dev-disk-by-label-nas btrfs rw,relatime,space_cache,subvolid=5,subvol=/ 0 0

      The disk is mounted several times? Oh, is this a bus-powered Seagate USB3 disk?

      Source Code

      1. [ 8.757964] scsi 0:0:0:0: Direct-Access Seagate Expansion 9300 PQ: 0 ANSI: 6
      If yes, I'm out (consider these things unreliable crap)

      kamilmirza wrote:

      benchmark the same HDD with NTFS it shows above 110 MB/s on Windows
      So you constantly reformat the disk and switch between Windows and Linux?

      Anyway, the numbers are as they are. I doubt that you run Windows on your ODROID-XU4 (so you're comparing different hardware setups) and for me personally it's not worth the time to diagnose problems with USB3 Seagate disks (or bus-powered disks connected to XU4 in general, I consider this board a problematic NAS choice unlike ODROID HC1 and HC2)

      Somewhere in this thread Building OMV automatically for a bunch of different ARM dev boards you'll find me testing with XU4 and a underpowered 2.5" disk that shows low performance. Too lazy to search for on my own...

      Edit: usually it's about undervoltage and I should really stop trying to help with crappy hardware setups since it's just a waste of time...
    • kamilmirza wrote:

      it used to be NTFS before
      Then I would suggest testing with another Linux host or reformatting with NTFS again and testing with Windows (and not with 2 USB hubs in between, the XU4 has already one internal hub). Anyway: poor storage write performance is the reason for poor NAS write performance and 'jitter'. A long SMART selftest is also a good idea.

      As a reference: Building OMV automatically for a bunch of different ARM dev boards (performance sucks when drive is undervolted)