Announcement

Collapse
No announcement yet.

Second hard drive, very slow write.

Collapse
This topic is closed.
X
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Second hard drive, very slow write.

    I've got a SSD primary HD and a terabyte HD for storage. The terabyte hard drive is running sloooooow. It's taking 20+ minutes to move 1.5 gigs from the SSD to the terabyte HD. It's impossible to save anything big to the terabyte HD from GIMP. It'd take forever if it didn't crash GIMP hard. Need this fixed.

    #2
    What file system is on the terabyte drive? Is it attached via SATA or USB 2.0 or what? Also, how big is each drive?
    I'd rather be locked out than locked in.

    Comment


      #3
      Originally posted by SecretCode View Post
      What file system is on the terabyte drive? Is it attached via SATA or USB 2.0 or what? Also, how big is each drive?
      Both are ext4, Both are attached to the MB via SATA, one is a 60GB SSD and the other is a TB HD.

      Comment


        #4
        OK, so changing to ext4 and connecting via SATA isn't going to make any difference

        What are the mount options for the second drive (from mount)? (Clutching at straws here...)

        While a copy is running, does top report any processes maxing CPU? Does sudo iotop show anything unusual?

        Have you fsck'd both drives?

        My usual suspects for slow data transfer, apart from the speed of the interface itself and the drivers involved, are caching on write (but I don't think that applies to SATA) ... or errors.
        I'd rather be locked out than locked in.

        Comment


          #5
          Originally posted by SecretCode View Post
          What are the mount options for the second drive (from mount)? (Clutching at straws here...)
          /dev/sdb1 on /media/manon type ext4 (rw,nosuid,nodev,sync,commit=0)

          Originally posted by SecretCode View Post
          While a copy is running, does top report any processes maxing CPU? Does sudo iotop show anything unusual?
          kio_thumbnail was 100% cpu for the first third of the copy, then it died. After that, Firefox and plasma-desktop used the most cpu resources, usually around the 2% mark but peaking to 11% (Firefox).

          Googling around, kio_thumbnail is causing all kinds of trouble. I haven't figured out how to turn it off in Dolphin though.

          The iotop command was "not found."

          Originally posted by SecretCode View Post
          Have you fsck'd both drives?
          Will look into it . . .

          Comment


            #6
            You can install iotop from the repos

            That kio_thumbnail utilisation sounds suspicious. Haven't seen such a process hogging CPU on my system.

            Are you doing these copies from Dolphin? Does it make any difference if you use cp or rsync? And, thinking about it, does it make any difference if you boot into text mode, without starting the GUI?
            I'd rather be locked out than locked in.

            Comment


              #7
              I turned off all the previews in Konqueror and used it to copy a 230 meg file from SSD to TB HD. It took 5 minutes, during which I kept my eyeballs glued to the output from top. Kio_thumbnail never showed up, nothing used more than 2% cpu (plasma_desktop and Firefox) or a couple megs of RAM (Firefox).

              The problem appears to be application independent. It's also slow saving from GIMP.

              Let me try fsck next.

              Comment


                #8
                Result of fsck on the TB HD -

                sudo fsck /dev/sdb1
                fsck from util-linux 2.19.1
                e2fsck 1.41.14 (22-Dec-2010)
                manon: clean, 703/61054976 files, 6177685/244190000 blocks
                It sure didn't take long to run.

                At first I got an error message about trying to run fsck on a mounted drive. I unmounted the TB HD, so, not a problem. How would I run fsck on the SSD? The OS is running from it, can it be unmounted?

                Comment


                  #9
                  Doing a big copy - sudo iotop shows 94-99% in the IO category for command [jbd2/sdb1-8]. Read and write speeds for this command are 0%.

                  Below that there is a disk read speed 500-750 K/s and write speed 600-800 K/s for command kdeinit4: ki~.slave-socket. IO for that command is between .5-1%, peaking to 2.44%

                  Comment


                    #10
                    What is the output of:
                    Code:
                    sudo hdparm -itT /dev/sdb

                    Comment


                      #11
                      /dev/sdb:

                      Model=ST31000524AS, FwRev=JC4B, SerialNo=6VPGWCVQ
                      Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
                      RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
                      BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
                      CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=1953525168
                      IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
                      PIO modes: pio0 pio1 pio2 pio3 pio4
                      DMA modes: mdma0 mdma1 mdma2
                      UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
                      AdvancedPM=no WriteCache=enabled
                      Drive conforms to: unknown: ATA/ATAPI-4,5,6,7

                      * signifies the current active mode

                      Timing cached reads: 8938 MB in 2.00 seconds = 4472.47 MB/sec
                      Timing buffered disk reads: 290 MB in 3.01 seconds = 96.50 MB/sec


                      Done in the middle of a save from GIMP.

                      Comment


                        #12
                        Originally posted by Jeremy_Ray View Post
                        /dev/sdb:

                        Model=ST31000524AS, FwRev=JC4B, SerialNo=6VPGWCVQ
                        Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }
                        RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
                        BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
                        CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=1953525168
                        IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
                        PIO modes: pio0 pio1 pio2 pio3 pio4
                        DMA modes: mdma0 mdma1 mdma2
                        UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
                        AdvancedPM=no WriteCache=enabled
                        Drive conforms to: unknown: ATA/ATAPI-4,5,6,7

                        * signifies the current active mode

                        Timing cached reads: 8938 MB in 2.00 seconds = 4472.47 MB/sec
                        Timing buffered disk reads: 290 MB in 3.01 seconds = 96.50 MB/sec
                        Nothing immediately apparent there (using udma and writecache is enabled).
                        I'll get back to you if I can think of something else to test.

                        Comment


                          #13
                          One other thing to try: do a copy (of a single large file) using rsync with the --progress parameter. Does the transfer speed stay fairly constant or does it oscillate between fast and slow? (I get about 22-25MB/s fairly steady, although on previous installations I've often had problems with transfers starting fast and then slowing to less than 3MB/s - and sometimes speeding up again randomly.)

                          How much RAM do you have?

                          The high io utilisation for [jbd2/sdb1-8] may be a suspect. This process or thread relates to ext4 journalling (a good thing to have, generally). I'm not sure what to do about it though ...


                          You might want to install smartmontools and run sudo smartctl -i /dev/sdb (or whatever the terabyte drive is).
                          I'd rather be locked out than locked in.

                          Comment


                            #14
                            Originally posted by SecretCode View Post
                            How much RAM do you have?
                            16 gigs with 20 gig swap partition on the SSD.

                            I'd look at the other stuff but I've got a 20 minute save going :eek:

                            Comment


                              #15
                              16GB? I'm not speaking to you any more :mad:
                              I'd rather be locked out than locked in.

                              Comment

                              Working...
                              X