raspbian jessie | Steinar Bangs blogg

At the end of Using a Raspberry Pi 2 Model B as a router/firewall for the home LAN I wrote that I decided not to put /var/log into tmpfs, because:

I wanted the logs to be persistent
I thought that the wear would result in less and less of the sd card to become available (and 16GB for logs should last a loong time)

As it turned out the sd card died after one month.

I don’t know if the cause was excessive logging, the use of ntopng (which did write quite a lot, both in the number of files, the number of files, and in the total storage used, which was approximately 0,5GB after 30 days of uptime) or simply a bad sd card.

However, going forward with a new sd card, I’ve done the following:

Removed ntopng
Put /var/log on tmpfs (limited to 100MB in size), synced to a backing store on the sd card using rsync

For setting up the logging I found some existing web pages that took me part of the way, but not all the way:

Raspberry Pi: Extending the life of the SD card (this is where I got the fstab line from)
Observium persistent ramdisk (this is where I got the script and the crontab settings from)
Setting up overlayFS on Raspberry Pi (this is the where I got the systemctl command to set up the init.d script, as well as the script headers to make it work)
How does systemd use /etc/init.d scripts? (more information about how systemd uses /init.d scripts)

Here is what I did:

Logged in as root and did everything below as root

Edited /etc/fstab and added the following line:

tmpfs    /var/log    tmpfs    defaults,noatime,nosuid,mode=0755,size=100m    0 0

Created an /etc/init.d/ramdiskvarlog file with the following contents

#!/bin/sh
### BEGIN INIT INFO
# Provides:          ramdiskvarlog
# Required-Start:    $local_fs $time
# X-Stop-After:      $time
# Required-Start:    $local_fs $time
# Required-Stop:     $local_fs
# Default-Start:     S
# Default-Stop:      0 1 6
# Short-Description: Restore to and save logs from tmpfs filesystem
# Description:       Restore to and save logs from tmpfs filesystem
### END INIT INFO

# /etc/init.d/ramdiskvarlog
#

case "$1" in
  start)
    echo "Copying files to ramdisk"
    rsync -av /var/backup/log/ /var/log/
    echo [`date +"%Y-%m-%d %H:%M"`] Ramdisk Synched from HD >> /var/log/ramdisk_sync.log
    ;;
  sync)
    echo "Synching files from ramdisk to Harddisk"
    echo [`date +"%Y-%m-%d %H:%M"`] Ramdisk Synched to HD >> /var/log/ramdisk_sync.log
    rsync -avy --delete --recursive --force /var/log/ /var/backup/log/
    ;;
  stop)
    echo "Synching logfiles from ramdisk to Harddisk"
    echo [`date +"%Y-%m-%d %H:%M"`] Ramdisk Synched to HD >> /var/log/ramdisk_sync.log
    rsync -av --delete --recursive --force /var/log/ /var/backup/log/
    ;;
  *)
    echo "Usage: /etc/init.d/ramdisk {start|stop|sync}"
    exit 1
    ;;
esac

exit 0

Made /etc/init.d/ramdiskvarlog executable:
```
chmod +x /etc/init.d/ramdiskvarlog
```
Created a directory to store the logs persistently, and populated it initially with the contents of the existing /var/log with the following command line commands :
```
mkdir -p /var/backup/log
/etc/init.d/ramdiskvarlog sync
```
Made the /etc/init.d/ramdiskvarlog script be run at boot time and during orderly shutdown with the following command line command
```
systemctl enable ramdiskvarlog
```
Made the /etc/init.d/ramdiskvarlog script copy the contents of /var/log to the sd card once every 24 hours
1. At the command line gave the command
```
crontab -e
```
2. In the editor that opened on the crontab, added a line with the following contents
```
2 7 * * * /etc/init.d/ramdiskvarlog sync >> /dev/null 2>&1
```
Created a test file with “touch /var/log/test.log”, rebooted the raspberry pi with “sync; reboot”, and then:
1. Checked with the mount command that /var/log was on tmpfs, found the following line in the output, which meant that /var/log was on tmpfs
```
tmpfs on /var/log type tmpfs (rw,nosuid,noatime,size=102400k,mode=755)
```
2. Checked that the /var/log/test.log file was present (and the file was present, which meant that it had been synced to persistent storage on shutdown and restored on boot)

After completing the setup, I popped the sd card out and put it into a card reader on a debian desktop computer. Then I made an image of the working sd card, so that if/when the sd card dies, getting a working router again should be as quick as just dd’ing the image to a new sd card and then switching sd card on the raspberry Pi.

Lesson learned!

Since 1999 I have been using a 1996 vintage DEC PII desktop as the router/firewall between the internet and my home network. The DEC computer came to me with Win95 (or possibly Win98) in 1998, got SuSE linux and started its mission as router and firewall (and CUPS server, and IMAP server, and various other server stuff). When upgrading the SuSE installation to a newer version went south, it spent a while running ThomasEz’s floppyfw, until I used a floppy net install to install debian potato, immediately switched it to debian testing, until debian woody arrived, when it was moved to debian stable, and then I just kept running “apt-get dist-upgrade” until I finally had it running debian 8 “jessie” on june 6 in 2015.

The old DEC desktop has survived its maker company, survived lightning strikes that have sent the power supplies and/or main boards of other computers on the same LAN into continously beeping mode (i.e. broken). However, in December 2015 it started acting up, and crashing with irregular intervals (sometimes two weeks, sometimes one day).

So… the time for a replacement would have to be not too far ahead. The question was what to replace it with?

The simplest solution would be to just get a wireless router with a cabled switch. But that would mean:

No possibilities for SSH or mosh into the home LAN
No ntop
No support for netboot and TFTP in the home LAN
Limited, cumbersome and inflexible firewall setup

My requirements were:

Cheap
Two wired NICs
The ability to run debian
Preferrably fanless
Compact

ThomasEz immediately suggested using a raspberry pi with two NICs, but I thought that would be too puny, and I investigated alternatives like Shuttle Barebone DS57U but I found that the raspberry pi alternative was so cheap, I might as well order one.

And then it turned out to be so simple to set up so I had it up and running before I really had decided on anything, so now the r-pi is what I have.

This is what I ordered:

Raspberry Pi 2 Model B Starter Kit
TP-Link UE300 USB 3.0 to GbE Adapter (it was listed as being supported out of the box on raspberry pi)

Here’s what I did:

Downloaded the Raspbian Jessie Lite image to a debian jessie computer and unpacked it into the /tmp directory
Plugged an USB SD card reader into the debian computer, and followed the instructions in Installing operating system images on Linux
I plugged the cheapest USB keyboard I could get from my local teknikmagasinet store into one of the USB port, yanked the HDMI cable from the DVD player and plugged the r-pi into the TV, plugged a network cable into the local LAN, and plugged in the power… and the raspberry pi booted quickly into the familiar debian login
I logged in with the built-in “pi” user with password “raspberry”, and created my own user with the following command line command:
```
adduser sb
```
the changed the password of the root user and removed the pi user
I copied in a public ssh keys from my other computers, and put them into the ~/.ssh/authorized_keys file and then opened /etc/ssh/sshd_conf in a text editor and modified it in the following way:
1. Disabled root login by changing
```
PermitRootLogin without-password
```
  to
```
PermitRootLogin no
```
2. Disabled password login by changing
```
#PasswordAuthentication yes
```
  to
```
PasswordAuthentication no
```
  (removed the comment and changed “yes” to “no”)
Edited /etc/hostname to change the name from the default “raspberrypi” to “ocon”
Rebooted the pi to check the startup state of the ssh daemon and ssh’d in

Resized the disk to fill the entire SD card:

Typed the command
```
raspi-config
```

Selected

1 Expand Filesystem            Ensures that all of the SD card storage is available to the OS

and got the response

Root partition has been resized.
The filesystem will be enlarged upon the next reboot

Rebooted the system to get the full 16GB in the file system

Updated the system by giving the following command line commands:
```
apt-get update
apt-get dist-upgrade
```
(the “update” command updates the local package database against the package servers. The “dist-upgrade” command upgrades all packages that have a newer version, and the required dependencies)
Installed some useful software:
1. GNU emacs (my favorite text editor)
```
apt-get install emacs
```
2. mosh
```
apt-get install mosh
```
3. git (I’ve got my home directory versioned in git)
```
apt-get install git
```
4. rcs (I use it to version control operating system configuration files)
```
apt-get install rcs
```
I cloned my home directory in git and created a new branch (I have a different branch for each computer)

I set the built-in NIC permanently as eth0:

export INTERFACE=eth0
export MATCHADDR=`ip addr show $INTERFACE | grep ether | awk '{print $2}'`
/lib/udev/write_net_rules

I added configuration for a second NIC by adding the following to /etc/network/interfaces:

# The internal network card
allow-hotplug eth1
iface eth1 inet static
   address 10.10.10.1
   netmask 255.255.255.0

I plugged in the USB NIC to have it appear, and then made the USB NIC permanently eth1 with the following command line commands:
```
export INTERFACE=eth1
export MATCHADDR=`ip addr show $INTERFACE | grep ether | awk '{print $2}'`
/lib/udev/write_net_rules
```
Installed dnsmasq
```
apt-get install dnsmasq
```
Edited /etc/dnsmasq.conf to make dnsmasq respond to DHCP requests on eth1:
1. Removed the comment in front of
```
#interface=
```
  and set “eth1” as the value:
```
interface=eth1
```
2. Uncommented the domain directive
```
#domain=thekelleys.org.uk
```
  and changed it to my domain
```
domain=hjemme.lan
```
3. Uncommented the dhcp-range directive
```
#dhcp-range=192.168.0.50,192.168.0.150,12h
```
  and changed it to a 10.10.10.* range with a 5h lease on the addresses
```
# Our HOME LAN 5h lease time
dhcp-range=10.10.10.6,10.10.10.40,5h
```
Opened the /etc/hosts file in a text editor and added the raspberry pi itself, to so that DNS lookups of the raspberry pi will work in a LAN where the raspberry pi is handling the DHCP requests (dnsmasq will handle DNS requests for the IP addresses it has given DHCP leases to, as well as what it finds in the hosts file. The rest is delegated to the upstream DNS server)
```
127.0.0.1       localhost
::1             localhost ip6-localhost ip6-loopback
ff02::1         ip6-allnodes
ff02::2         ip6-allrouters

127.0.1.1       ocon

# local hosts
10.10.10.1  hjemme ocon hjemme.hjemme.lan ocon.hjemme.lan
```
Edited the /etc/sysctl.conf file to set up IPv4 routing in the linux kernel, removed the comment in front of the net.ipv4.ip_forward line:
```
# Uncomment the next line to enable packet forwarding for IPv4
net.ipv4.ip_forward=1
```

ferm is a utility that makes it easy to set the routing and firewall rules at boot time

Installed ferm using apt-get from a command line:
```
apt-get install ferm
```

Modified the /etc/ferm/ferm.conf file to allow everything inside t oroute out, but only allow ssh in

@def $DEV_WORLD = eth0;
@def $DEV_PRIVATE = eth1;

def $NET_PRIVATE = 10.10.10.0/24;

table filter {
    chain INPUT {
        policy DROP;

        # connection tracking
        mod state state INVALID DROP;
        mod state state (ESTABLISHED RELATED) ACCEPT;

        # allow local packet
        interface lo ACCEPT;

        # allow private net
        interface $DEV_PRIVATE ACCEPT;

        # respond to ping
        proto icmp ACCEPT;

        # allow IPsec
        proto udp dport 500 ACCEPT;
        proto (esp ah) ACCEPT;

        # allow SSH connections
        proto tcp dport ssh ACCEPT;
    }
    chain OUTPUT {
        policy ACCEPT;

        # connection tracking
        #mod state state INVALID DROP;
        mod state state (ESTABLISHED RELATED) ACCEPT;
    }

    chain FORWARD {
        policy DROP;

        # connection tracking
        mod state state INVALID DROP;
        mod state state (ESTABLISHED RELATED) ACCEPT;

        # connections from the internal net to the internet or
        # to other internal nets are allowed
        interface $DEV_PRIVATE ACCEPT;

        # the rest is dropped by the above policy
    }
}

table nat {
    chain POSTROUTING {
        # masquerade private IP addresses
        saddr $NET_PRIVATE outerface $DEV_WORLD MASQUERADE;
    }
}

The version of ferm in “jessie” doesn’t start at boot, because “jessie” dropped SYSV init in favour of systemd, and the version of ferm in “jessie” doesn’t have a systemd configuration, so I needed to manually download and install the version of ferm from debian testing (I downloaded from regular debian, since ferm doesn’t have anything platform specific):
```
cd /tmp
wget http://ftp.no.debian.org/debian/pool/main/f/ferm/ferm_2.2-5_all.deb
dpkg --install /tmp/ferm_2.2-5_all.deb
```
fail2ban monitors log files of daemons and adjust the firewall rules to temporary ban hosts it suspects of intrusion attempts. The debian (and raspbian) version of fail2ban will out of the box scan the logs for ssh intrusion attempts, so no configuration is necessary
To have an easy way of monitoring the network traffic in and out of the home LAN, I installed ntop ng
```
apt-get install ntopng
```
after the installation it is possible to monitor the network traffic by accessing http://ocon.hjemme.lan:3000 (the interesting traffic will be seen after selecting eth1)
The Network Time Protocol is how computers stay in sync, installing the ntp package will make the gateway keep network time, a
```
apt-get install ntp
```
Opened the /etc/ntp.conf file in a text editor, and modified it to provide an NTP deamon for the home LAN, uncommented the “broadcast” line and modified the network match to match the 10.10.10.* network:
```
# If you want to provide time to your local subnet, change the next line.
# (Again, the address is an example only.)
broadcast 10.10.10.255
```
Installed the apticron utility to make sure that the APT database is updated daily with new candidates for update
```
apt-get install apticron
```

The original plan was to run the raspberry pi headless, but since I had an old VGA only LCD display for the old DEC computer I might as well hook it up the raspberry pi, together with the cheap USB keyboard used for setup.

I bought an HDMI to VGA converter with the manufacturer id VLMP34900W0.20. I plugged it in between the display and the raspberry-pi the display stayed black. I edited the /boot/config.txt file, removing the comment in front of the hdmi_safe line:

# uncomment if you get no picture on HDMI for a default "safe" mode
hdmi_safe=1

I rebooted the raspberry pi, and this time the LCD displayed showed the boot messages as well as a normal console login prompt.

The raspberry pi 2 model B, with an extra USB NIC, a USB keyboard and connected to a VGA display using an HDMI to VGA converter

And this is where the current state is. One initial concern was flash wear on the SD card, which doesn’t have the wear leveling features of a “real” SSD, so I had some plans on making the /var/log use tmpfs.

But I decided not to, since having real persistent logs is a useful thing for a gateway, and since 16GB is actually an awful lot of data if all you do is to write textual files. And ff the SD card wears out I’ll just by a new SD card, and make a new system. Since I now know how, this shouldn’t take long

Steinar Bangs blogg

Tag Archives: raspbian jessie

Logging to persistent tmpfs on Raspbian “jessie”

Using a Raspberry Pi 2 Model B as a router/firewall for the home LAN

On Emacs, eclipse, and other things that take my fancy