Mar 082007

Jails under FreeBSD 6

This article shows you how I created a jail for under FreeBSD 6. This article is based on a previous article. I have written previously about jails on FreeBSD 4 and on FreeBSD 5 . The goal of this jail is the creation of a test environment for updating the Bacula ports that I maintain

Lift and separate

Jails can be used to separate different processes and keep them apart from each other so they cannot interfere. For example, you could run Apache in a jail and keep it away from everything else on the machine. Should an exploit be found in Apache and used to compromise your system, the intruders can only do what the jail allows them to do. A jail can consist of a full operating system, or a single executable. From within a jail, they are chroot‘d and cannot see anything outside of the jail. At the same time, it appears to them as if they are running on their own machine with their own operating system. As far as they know, they have their own computer and nobody else is on the system. Running a virtual system within a jail is a good solution if you want to provide someone with resources, but don’t want them to have complete control over your system. A jail can help you deal with issues of security, access, and increase utilization of existing resources, all at the same time.

Jail documentation

The main document for creating a jail is man jail. I followed the instructions listed under Setting up a Jail Directory Tree. I used those instructions to create the jail. You will need the full source tree for the system you going to create. I used the /usr/src/ directory I had from my most recent build world. One step from man jail that I did not follow:
  • I left sendmail (actually, postfix) running. I just changed it so that it did not listen on all IP addresses. I added this setting to /usr/local/etc/postfix/
    inet_interfaces = $myhostname
    That setting allows the jail to run its own mail server.
I put my jail at /home/jails/ducky/. This is the value I assigned to D in the instructions. After you have installed the jail, if you peek inside that directory, you’ll see it looks just like the root directory of a typical FreeBSD system:
[dan@dfc:/home/jails/ducky/] $ ls
COPYRIGHT       etc             libexec         root            usr
bin             home            mnt             sbin            var
boot            kernel          proc            sys
dev             lib             rescue          tmp
[dan@dfc:/home/jails/ducky/] $

Terminology: host versus jail

The host environment is the main system and is where you first install FreeBSD on the computer. It is in the host environment that you create a jail. The Bacula project will do their testing in the jail. They have access to the jail and only the jail. They will not have access to the host environment at all. This concept of host environment and jail environment will be used later in this article. It is important that you understand what each one is. In this example, the host environment will be at IP address and the jail will be at

Modifying other daemons

Most daemons will listen to whatever IP addresses are available to them. After starting your jail, if you try to ssh to it, you will not get into it. You’ll be in the host environment instead. To get into the jail environment via ssh, you need to:

  • Tell the host environment sshd not to listen to the jail’s IP address
  • run sshd in the jail

Host environment syslogd

This entry in /etc/rc.conf tells syslogd to not listen on any IP address.

That allows syslogd to run in both the host and the jail environments.

Host environment inetd

This entry in /etc/rc.conf tells inetd to listen on a specific IP address. This address is that of the host environment:

inetd_flags="-wW -C 60 -a"

You should note that the first part of the above flags is from /etc/defaults/rc.conf:

inetd_flags="-wW -C 60" # Optional flags to inetd

Host environment sshd

To alter the host environment sshd so it listens only to host environment IP addresses, modify /etc/ssh/sshd_config and set the IP address for the Listen directive:
Then restart the main sshd process:
kill -HUP `cat /var/run/`
Use telnet to verify that the host environment is not listening on the jail address:
$ telnet 22
telnet: connect to address Connection refused
telnet: Unable to connect to remote host
If you don’t get a connection, the host environment is not listening. This assumes that you have not yet started sshd in the jail environment.

Jail environment sshd

To start sshd in the jail environment, add the following line to /etc/rc.conf:


To get DNS working, add something like this to /etc/resolv.conf:


Jail environment syslogd

In addition, I also swapped console output to /var/log/messages as shown in this snipped from /etc/syslog.conf:
#*.err;kern.warning;auth.notice;mail.crit               /dev/console *.err;kern.warning;auth.notice;mail.crit                /var/log/messages

Configuring the Jail

Next, you’ll want to read the part of the man page titled Configuring the Jail. Included there you will to make the following settings within the jail. I made these changes to the jail directly from the host environment (that is, I did not start the jail; I modified the files from outside). These are things I changed, but I can’t point to a man page as to why it’s a good thing to do this:
  • adjkerntz – Not sure about this. I commented out the /etc/crontab entry for adjkerntz within the jail environment. If you don’t do this, you’ll see this type of notification from cron via email:
    adjkerntz[11643]: sysctl(put_wallclock): Operation not permitted
  • /etc/ssh – I was actually duplicating an existing physical machine into this environment. Therefore, copying over the keys from this directory will avoid “WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!” warning messages. If you are creating a new environment, you don’t have to worry about this step.

Starting the jail for the first time

From man jail, to start a jail, issue this command:
[root@mtwenty:/home/dan] # jail /home/jails/ducky/ /bin/sh
That prompt (#) indicates you are now in the jail environment. Now you can run the start up processes:
# sh /etc/rc
Loading configuration files.
Setting hostname:
Generating nsswitch.conf.
Generating host.conf.
Starting syslogd.
ELF ldconfig path: /lib /usr/lib /usr/lib/compat /usr/local/lib
a.out ldconfig path: /usr/lib/aout /usr/lib/compat/aout
Starting local daemons:.
Updating motd.
/etc/rc: WARNING: Setting entropy source to blocking mode.
Type a full screenful of random junk to unblock
it and remember to finish with . This will
timeout in 300 seconds, but waiting for
the timeout without typing junk may make the
entropy source deliver predictable output.

Just hit for fast+insecure startup.
kern.random.sys.seeded: 1
jalkjlkajdflkajdfl iur opiquv dropuivwaopieuaoijdfl;uiop9^[[12~84718e0r9invpfinadfpisad;ifsda;lsajdfl lk;kasf;kladfs
Generating public/private rsa1 key pair.
Your identification has been saved in /etc/ssh/ssh_host_key.
Your public key has been saved in /etc/ssh/
The key fingerprint is:
Generating public/private dsa key pair.
Your identification has been saved in /etc/ssh/ssh_host_dsa_key.
Your public key has been saved in /etc/ssh/
The key fingerprint is:
Generating public/private rsa key pair.
Your identification has been saved in /etc/ssh/ssh_host_rsa_key.
Your public key has been saved in /etc/ssh/
The key fingerprint is:
Starting sshd.
Starting cron.
Local package initialization:.

Sun Sep 11 17:22:42 EDT 2005
For the most part, this looks exactly like a normal startup. While you have it running, you might want to add a user, set the root password, etc. I had some problems with ps:
# ps auwx
ps: bad namelist
This usually indicate a kernel that is not in sync with world. To fix this problem, I repeated some of the steps under man 8 jail.
make distribution DESTDIR=$D
mount_devfs devfs $D/dev
cd $D
ln -sf dev/null kernel
In hindsight, I think I missed the mount_devfs step. Symptoms included getting logged in by ssh, but then the screen would freeze. The above steps fixed that problem.

Starting and stopping the jail automagically

With 6.2 comes /etc/rc.d/jail, a startup script for jails. With a bit of fishing in /etc/defaults/rc.conf and some trial and error, I figured out how to use this script. Here are my entries from /etc/rc.conf:

# Defaults for all jails:
jail_interface="fxp0"    # Interface to create the IP alias on
jail_devfs_enable="YES"  # mount devfs in the jail
jail_procfs_enable="YES" # mount procfs in jail

# list of jails on this machine

# values for each jail listed above
jail_ducky_rootdir="/usr/home/jails/ducky" # Jail's root directory
jail_ducky_hostname=""    # Jail's hostname
jail_ducky_ip=""              # Jail's IP number
I could have used jail_ducky_interface, jail_ducky_devfs_enable, and jail_ducky_procfs_enable, but all jail variables, except rootdir, hostname and ip, can be defined without the jail name. Thus, you can set defaults, and then specify only the exception. A fine FreeBSD tradition.

Additional resources

Just after this article appeared on ONLamp, Anthony Nguyen wrote in with these great resources:
  • FreeBSD Jails by Khairil Yusof, Tom McLaughlin, Radek Kozlowski.
He also mentioned:
  1. Before you even set up a jail if you have already done a build world, you can skip a compiling step. quote “If you have already built world, you can replace make world with make installworld on the fourth line to avoid recompiling everything again.”
  2. After you set up your first jail and want more jails, look for the heading “Reusing jails”. It clones it.
  3. Starting jail automatically, look at heading “Using system rc scripts to start and stop jails”.

Paul Dekkers also wrote in with:
You know you don’t need jailtools anymore these days to have rc-scripts and all, don’t you? You can just use the right parameters in /etc/rc.conf, like:

(Which is real example ;-)) Starting and stopping then goes with /etc/rc.d/jail start and I believe you can add the jail name after that. |BTW, I never really needed the security.jail.set_hostname_allowed because I set the hostname in the rc.conf command|

Jails run well

Jails run virtual machines very well. They look very much like a real system. You must look pretty close to be able to tell you’re in a jail. My jail allows the Bacula developers to have a machine of their own. It also allows me to keep their work totally separate from my own. A jail can be used to deal with security issues and to increase the utilization of an existing machine while giving everyone their own virtual machine. There’s no reason why you couldn’t run many different jails on the same computer. Enjoy.

  2 Responses to “Jails under FreeBSD 6”

  1. The following patch fixes the adjkerntz issue:


    And the following patch allows you to set the netmask of jails.


    Hope they’re useful to someone. -sc