Guide: NAS using OpenMediaVault (like BanaNAS but based on Bananian)

angeloD

I wanted to build a NAS using my BananaPi, I tried BanaNAS but it did not boot correctly on my one, additionally the OpenMediaVault guys strongly advise NOT to run it on a flash drive (because of the multiple writes that, in the long run will destroy the medium).

So I decided to make one myself based on Bananian (BanaNAS is based on Raspian lift), addressing (or at least trying to address) at the same time the wearing of the flash drive. This is how I made it:

1. After the boot find the ip of the machine

1. ifconfig -a

Copy the Code

or for headless installs, on a machine on the same network

1. nmap -n -sP 192.168.0.0/24

Copy the Code

2. ssh to the ip found: root@ip, password pi

3. run bananian-config

4. To switch off the blinking green led, add the line:

1. echo none > /sys/class/leds/green\:ph24\:led1/trigger

Copy the Code

to the /etc/rc.local file.

5. Set a fixed ip, edit the file /etc/network/interfaces

1. # static ip configuration
   
2. iface eth0 inet static
   
3.         address 192.168.0.199
   
4.         netmask 255.255.255.0
   
5.         gateway 192.168.0.1

Copy the Code

6. install openmediavault

1. echo "deb http://packages.openmediavault.org/public kralizec main" > /etc/apt/sources.list.d/openmediavault.list
   
2. apt-get update
   
3. apt-get install openmediavault-keyring postfix
   
4. apt-get update
   
5. apt-get install php-apc
   
6. apt-get install openmediavault
   
7. omv-initsystem
   
8. 
   
9. shutdown -r now

Copy the Code

7. ssh is now disabled, access the web interface, from a browser

User: admin
PW: openmediavault

8. Add DNS server in System/Network/DNS Server to allow name resolution

9. add extras repository

1. echo "deb http://packages.omv-extras.org/debian/ kralizec main" >> /etc/apt/sources.list.d/omv-extras-org-kralizec.list
   
2. apt-get update
   
3. apt-get install openmediavault-omvextrasorg
   
4. apt-get update

Copy the Code

10. Generate, in OMV a certificate to enable https.

11. Switch off the usage of swap filesystem

1. swapoff -a

Copy the Code

12. Switch off journaling in root filesystem, remove the SD from the banana, from another machine

1. 
   
2. tune4fs -O ^has_journal /dev/...

Copy the Code

13. Use /var/log in tmpfs (taken from https://www.debian-administratio ... _transient_/var/log)

Create the file /etc/init.d/transientlog with the following content:

1. #!/bin/sh
   
2. ### BEGIN INIT INFO
   
3. # Provides: transientlog
   
4. # Provides: transientlog
   
5. # Required-Start:
   
6. # Required-Stop:
   
7. # X-Start-Before: $syslog
   
8. # X-Stop-After: $syslog
   
9. # X-Interactive: yes
   
10. # Default-Start:        2 3 4 5
    
11. # Default-Stop:                0 1 6
    
12. # Short-Description:        Keeps /var/log in RAM
    
13. # Description: Moves the contents of /var/log to RAM during boot
    
14. #              and keeps it there until shutdown/reboot, when it
    
15. #              copies the contents back to permanent storage.
    
16. ### END INIT INFO
    
17. 
    
18. # Version 1.0
    
19. # Author: Matteo Cortese <matteo_cortese@fastwebnet.it>
    
20. #
    
21. # Inspired by ramlog by Jan Andrejkovic.
    
22. #
    
23. # This program is free software; you can redistribute it and/or modify
    
24. # it under the terms of the GNU General Public License as published by
    
25. # the Free Software Foundation; either version 2 of the License, or
    
26. # (at your option) any later version.
    
27. 
    
28. PATH=/sbin:/bin:/usr/sbin:/usr/bin
    
29. NAME=transientlog
    
30. DESC="Transient log directory"
    
31. LOCKFILE=/var/lock/$NAME.lock
    
32. VARLOG=/var/log
    
33. SIZE=16M
    
34. MODE=0755
    
35. 
    
36. # Read configuration variable file if it is present
    
37. [ -r /etc/default/$NAME ] && . /etc/default/$NAME
    
38. 
    
39. # Load the VERBOSE setting and other rcS variables
    
40. . /lib/init/vars.sh
    
41. 
    
42. # Define LSB log_* functions.
    
43. # Depend on lsb-base (>= 3.0-6) to ensure that this file is present.
    
44. . /lib/lsb/init-functions
    
45. 
    
46. if [ -z "$VARLOGPERM" ]; then
    
47.         var=`dirname $VARLOG`
    
48.         log=`basename $VARLOG`
    
49.         VARLOGPERM="$var/permanent$log"
    
50.         unset var log
    
51. fi
    
52. 
    
53. do_start()
    
54. {
    
55.         # Return
    
56.         #   0 if transient log has been started
    
57.         #   1 if transient log was already running
    
58.         #   2 if transient log could not be started
    
59. 
    
60.         [ -f $LOCKFILE ] && return 1
    
61.        
    
62.         # Check if I'm root
    
63.         [ `id -u` -eq 0 ] || return 2
    
64. 
    
65.         # If VARLOG does not exist?
    
66.         [ -d $VARLOG ] || return 2
    
67. 
    
68.         # VARLOGPERM either does not exist (first invocation)
    
69.         # or is empty (left from previous invocation).
    
70.         #
    
71.         [ -d $VARLOGPERM ] || mkdir -p $VARLOGPERM || return 2
    
72. 
    
73.         # Mount a tmpfs over VARLOG.
    
74.         # The mount will shadow the current contents of VARLOG.
    
75.         # So, before, make a bind mount so that looking into VARLOGPERM
    
76.         # we'll see the current contents of VARLOG, which
    
77.         # will not be available anymore as soon as we mount
    
78.         # a tmpfs over it.
    
79.         #
    
80.         mount --bind $VARLOG $VARLOGPERM
    
81.         # NEEDED FOR DEBIAN Jessie
    
82.         mount --make-private $VARLOGPERM
    
83.         mount -t tmpfs -o nosuid,noexec,nodev,mode=$MODE,size=$SIZE $NAME $VARLOG
    
84.         if [ $? -eq 0 ]; then
    
85.                 # Populate the tmpfs
    
86.                 if cp -rfp $VARLOGPERM -T $VARLOG; then
    
87.                         # Success!
    
88.                         touch $LOCKFILE
    
89.                         return 0
    
90.                 fi
    
91. 
    
92.                 # Something went wrong...
    
93. 
    
94.                 # Rollback the mount
    
95.                 umount -l $VARLOG
    
96.         fi
    
97. 
    
98.         # Rollback the directory mangling
    
99.         umount $VARLOGPERM
    
100.        
     
101.         return 2
     
102. }
     
103. 
     
104. do_stop() {
     
105.         # Return
     
106.         #   0 if daemon has been stopped
     
107.         #   1 if daemon was already stopped
     
108.         #   2 if daemon could not be stopped
     
109.         #   other if a failure occurred
     
110. 
     
111.         [ -f $LOCKFILE ] || return 1
     
112. 
     
113.         # Check if I am root
     
114.         [ `id -u` -eq 0 ] || return 2
     
115. 
     
116.         # Merge back to permanent storage
     
117.         cp -rfup $VARLOG -T $VARLOGPERM
     
118. 
     
119.         # The following cannot fail... or can it?
     
120.         umount -l $VARLOG
     
121.         umount -l $VARLOGPERM
     
122.         rm -f $LOCKFILE
     
123.         return 0
     
124. }
     
125. 
     
126. do_reload() {
     
127.         # Return
     
128.         #   0 if transient log has been reloaded
     
129.         #   1 if transient log was not running
     
130.         #   2 if transient log could not be reloaded
     
131. 
     
132.         [ -f $LOCKFILE ] || return 1
     
133.        
     
134.         # Check if I am root
     
135.         [ `id -u` -eq 0 ] || return 2
     
136. 
     
137.         # Merge back to permanent storage
     
138.         cp -rfup $VARLOG -T $VARLOGPERM
     
139.         touch $LOCKFILE
     
140.         return 0
     
141. }
     
142.        
     
143. case "$1" in
     
144.   start)
     
145.         [ "$VERBOSE" != no ] && log_daemon_msg "Starting $DESC" "$NAME"
     
146.         do_start
     
147.         case "$?" in
     
148.                 0|1) [ "$VERBOSE" != no ] && log_end_msg 0 ;;
     
149.                 2) [ "$VERBOSE" != no ] && log_end_msg 1 ;;
     
150.         esac
     
151.         ;;
     
152.   stop)
     
153.         [ "$VERBOSE" != no ] && log_daemon_msg "Stopping $DESC" "$NAME"
     
154.         do_stop
     
155.         case "$?" in
     
156.                 0|1) [ "$VERBOSE" != no ] && log_end_msg 0 ;;
     
157.                 2) [ "$VERBOSE" != no ] && log_end_msg 1 ;;
     
158.         esac
     
159.         ;;
     
160.   status)
     
161.           if [ -f $LOCKFILE ]; then
     
162.                 echo "$DESC is running"
     
163.         else
     
164.                 echo "$DESC is stopped"
     
165.                 exit 1
     
166.         fi
     
167.         ;;
     
168.   reload)
     
169.         log_daemon_msg "Reloading $DESC" "$NAME"
     
170.         do_reload
     
171.         log_end_msg $?
     
172.         ;;
     
173.   *)
     
174.         echo "Usage: $0 {start|stop|status|reload}" >&2
     
175.         exit 3
     
176.         ;;
     
177. esac
     
178. 
     
179. 
     

Copy the Code

Make it executable and insert as service:

1. chmod a+x /etc/init.d/transientlog
   
2. insserv transientlog

Copy the Code

As suggested (and indicated in the original page), optionally create a file "transientlogCron"
with this content:

1. #!/bin/sh
   
2. 
   
3. [ -x /etc/init.d/transientlog ] || exit 0
   
4. [ -f /var/lock/transientlog.lock ] || exit 0
   
5. /etc/init.d/transientlog reload

Copy the Code

make it executable, chmod +x and move it in the /etc/cron.daily folder.

Enjoy now your OMV NAS....

tkaiser

Shouldn't a cronjob be executed at least daily that fires up 'service transientlog reload' to let the changes be written to the SD card?

BTW: This transientlog method is a nice approach even if you moved your rootfs to a connected SATA disk since normally you would like the disk to be sleeping most of the times and prevent spin-ups due to log activity. On my ARM based backup NAS systems I put the following into /etc/hdparm.conf:

1. /dev/sda {
   
2.         spindown_time = 120
   
3. }

Copy the Code

angeloD

tkaiser replied at Tue Nov 18, 2014 03:45
Shouldn't a cronjob be executed at least daily that fires up 'service transientlog reload' to let th ...

Thanks tkaiser , I added the part you suggested.

For the spin down of the HD, in my case with a Western Digital Green (4Tb connected through a cradle ORICO 6518sus3), the default settings are pretty ok.

All comments and suggestions are more than welcome.

angeloD

To reduce the writes to the bare minimum it is also possible to stop the collectd and rrdcached tasks (if you are not interested in what they do...)

/etc/init.d/rrdcached stop
/etc/init.d/collectd stop
monit stop rrdcached
monit stop collectd

On Ext4 partitions, to monitor the amount of data written, an interesting file is:

/sys/fs/ext4/<<<your partition here>>>/session_write_kbytes

karlkloss

I want to install Bananian and OMV on the harddisk. Could I use the above method to avoid frequent writes to the disk so that it can spin down when nobody accesses it over the network?

tkaiser

karlkloss replied at Sat Jan 24, 2015 12:05
I want to install Bananian and OMV on the harddisk. Could I use the above method to avoid frequent w ...

This might work but you should be aware that there exist different locations elsewhere on the disk where processes modify data. It's not just /var/log.

Both iotop and iostat (part of the sysstat package) might help to get processes writing constantly to disk. If you really want the disk to spin down then keep the system on the SD card and use data/mountpoint of the disk only for NAS purposes.

angeloD

You should also be aware that, by default, you cannot use the disk where OMV is installed to store your files.

blindpet

Does bananian now work by default with port multipliers like the jm321? I plan to install OMV on a small SSD and run snapraid on 3 x 2 TB disks

tkaiser

blindpet replied at Fri Mar 20, 2015 16:55
Does bananian now work by default with port multipliers like the jm321?

Will be the case when Bananian starts to use mainline kernel and uses CONFIG_AHCI_SUNXI=m (then it works as outlined here). Currently (that means using any kernel prior to 3.18) it's not possible without both patching kernel sources and recompiling the kernel as outlined in the same thread.

And regarding any sort of RAID behind a cheap and unreliable JMB321: Be aware that this piece of silicone might fail in situations when you don't expect it (eg one disk becomes unavailable and the PM decides to drop all other disks too): http://forum.lemaker.org/forum.php?mod=redirect&goto=findpost&ptid=11857&pid=71617

blindpet

tkaiser replied at Sat Mar 21, 2015 03:33
Will be the case when Bananian starts to use mainline kernel and uses CONFIG_AHCI_SUNXI=m (then it ...

I have read a lot of your posts tkaiser, very informative. I understand that if one drive fails then everything gets fucked up when you are running a software raid. Is this still true for something like snapraid? What if I just use each drive separately as ext4 volumes, if a drive dies then it is an issue?

My goal is to move my media downloading to the Banana Pi and have 2 separate drives connected to the port mulitplier. I will run backup onto USB drives weekly.

I am curious if one drive fails without RAID if the data set will be compromised?

Thanks for all the time you put into posting your detailed findings.