Debian – Backup data rate on Raspberry Pi maxing out at 5 Mb/s. Why

afpdata transferdebianraspberry piusb-storage

I set up my Raspberry Pi as a Time Machine, as documented here.

At the moment, the Raspberry Pi is connected to my MacBook Pro using a direct Ethernet cable. Also, an external hard drive (laptop drive) is connected to the Raspberry Pi using the USB port. However, backups are pretty slow. Activity Monitor claims that the Network is transferring a very steady 5 Mb/s, where my Time Capsule is transferring up to 8 Mb/s with a lot of fluctuation.

The Raspberry Pi self-reports (top) that its CPU is only half-used, with about equal parts afpd, usb-storage and jbd2/sda1-8. Thus, I think that the processing power of the Raspberry Pi does not seem to be the problem here.

To me, this looks like there is some kind of bottleneck that maxes out at 5 Mb/s thus potentially having my backups run at less than their potential speed. To the best of my knowledge, this might be the afp-daemon, the usb-bus or the external hard drive.

So, my question is, how could I identify the true culprit and what can I do about it?

Best Answer

Raspberry Pi network card is connected via USB bus, so it's throughput is limited. 10MB/s [usb to disk + lan to usb] + 1-2MB/s for extra upstream data ~= 11.5MB/s which is fairly good result for device like RPi.

p.s. I think you talking about MB/s (megabytes per seconds) - not Mb/s (megabits per second).

Related Solutions

Linux – Force spin-down of external hard-drive on linux (raspberry pi)

I didn't have luck with hd-idle; it ran but didn't function. I ended up writing the script below:

#!/bin/bash
# This script looks for recent disk access, and if nothing has changed, puts /dev/"drive" into spindown mode.
# This should be used only is the hdparm power management function is not working.
# Call this script with cron or manually as desired
#
#
#
# Change which drive this script looks at by changing the drive variable below:
drive="sda"
#
#
current=`date`
caller=$(ps ax | grep "^ *$PPID" | awk '{print $NF}')
filename="/tmp/diskaccess.txt"
if [ -f "$filename" ]; then
    stat_old=`cat "$filename" | tr -dc "[:digit:]"`
    stat_new=`cat /sys/block/"$drive"/stat | tr -dc "[:digit:]"`
    if [ "$stat_old" == "$stat_new" ]; then
        stat="0"
        echo "The disk hasn't been used; spinning down /dev/$drive"
        echo $stat_old
        hdparm -y /dev/$drive > /dev/null
    else
        stat="1"
        echo $stat_old
        echo $stat_new
        echo "The drive has been used..."
        echo $stat_new > $filename
    fi
else
    echo "/tmp/diskaccess.txt file does not exist; creating it now."
    echo $stat_new > $filename
fi
echo $stat " - " $drive " - " $current " - by: " $caller >> /tmp/diskaccesslog.txt

Format an external drive with ext4

To create the partition I used fdisk by invoking sudo fdisk /dev/sda the interactive session is easy to follow :

Command (m for help): d 

Selected partition 1 Partition 1 has been deleted.

Command (m for help): n 

Partition number (1-128, default 1): First sector (34-1953525134, default 2048): 34 Last sector, +sectors or
+size{K,M,G,T,P} (34-1953525134, default 1953525134):


Command (m for help): w 

The partition table has been altered. Calling ioctl() to re-read partition table. Syncing disks.

Next, I formatted the partition:

pi@retina:~ $ sudo mkfs.ext4 /dev/sda1 

mke2fs 1.42.12 (29-Aug-2014) Creating filesystem with 244190637 4k blocks and 61054976 inodes Filesystem UUID: fe042d87-3266-44b5-beff-a98f70b783ed Superblock backups stored on blocks:    32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,     4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968,     102400000, 214990848

Allocating group tables: done Writing inode tables: done Creating journal (32768 blocks): done Writing superblocks and filesystem accounting information: done

Created a new partition 1 of type 'Linux filesystem' and of size 931.5 GiB.

finally I ejected the drive:

sudo eject /dev/sda

to re-connect it, such that it appears now in the mount table.

Related Question