sshd(8) has a very nice feature that is often overlooked. That feature is the ability to allow a ssh user to run a specified command and nothing else, not even a login shell.

This is really easy to use and the magic happens in the file authorized_keys which can be used to restrict commands per public key.

For example, if you want to allow someone to run the “uptime” command on your server, you can create a user account for that person, with no password so the password login will be disabled, and add his/her ssh public key in ~/.ssh/authorized_keys of that new user, with the following content.

restrict,command="/usr/bin/uptime"ssh-rsa the_key_content_here

The user will not be able to log-in, and doing the command ssh remoteserver will return the output of uptime. There is no way to escape this.

While running uptime is not really helpful, this can be used for a much more interesting use case, like allowing remote users to use vmctl without giving a shell account. The vmctl command requires parameters, the configuration will be slightly different.

restrict,pty,command="/usr/sbin/vmctl $SSH_ORIGINAL_COMMAND" ssh-rsa the_key_content_here"

The variable SSH_ORIGINAL_COMMAND contains the value of what is passed as parameter to ssh. The pty keyword also make an appearance, that will be explained later.

If the user connects to ssh, vmctl with no parameter will be output.

$ ssh remotehost
usage:  vmctl [-v] command [arg ...]
    vmctl console id
    vmctl create "path" [-b base] [-i disk] [-s size]
    vmctl load "path"
    vmctl log [verbose|brief]
    vmctl reload
    vmctl reset [all|vms|switches]
    vmctl show [id]
    vmctl start "name" [-Lc] [-b image] [-r image] [-m size]
            [-n switch] [-i count] [-d disk]* [-t name]
    vmctl status [id]
    vmctl stop [id|-a] [-fw]
    vmctl pause id
    vmctl unpause id
    vmctl send id
    vmctl receive id

If you pass parameters to ssh, it will be passed to vmctl.

$ ssh remotehost show
   ID   PID VCPUS  MAXMEM  CURMEM     TTY        OWNER NAME
1     -     1    1.0G       -       -       solene test
$ ssh remotehost start test
vmctl: started vm 1 successfully, tty /dev/ttyp9
$ ssh -t remotehost console test
(I)nstall, (U)pgrade, (A)utoinstall or (S)hell?

The ssh connections become a call to vmctl and ssh parameters become vmctl parameters.

Note that in the last example, I use “ssh -t”, this is so to force allocation of a pseudo tty device. This is required for vmctl console to get a fully working console. The keyword restrict does not allow pty allocation, that is why we have to add pty after restrict, to allow it.

In this article I will present you the rcs tools and we will use it for versioning files in /etc to track changes between editions. These tools are part of the OpenBSD base install.

Prerequisites

You need to create a RCS folder where your files are, so the files versions will be saved in it. I will use /etc in the examples, you can adapt to your needs.

# cd /etc
# mkdir RCS

The following examples use the command ci -u. This will be explained later why so.

Tracking a file

We need to add a file to the RCS directory so we can track its revisions. Each time we will proceed, we will create a new revision of the file which contain the whole file at that point of time. This will allow us to see changes between revisions, and the date of each revision (and some others informations).

I really recommend to track the files you edit in your system, or even configuration file in your user directory.

In next example, we will create the first revision of our file with ci, and we will have to write some message about it, like what is doing that file. Once we write the message, we need to validate with a single dot on the line.

# cd /etc
# ci -u fstab
fstab,v  <--  fstab
enter description, terminated with single '.' or end of file:
NOTE: This is NOT the log message!
>> this is the /etc/fstab file
>> .
initial revision: 1.1
done

Editing a file

The process of edition has multiples steps, using ci and co:

1. checkout the file and lock it, this will make the file available for writing and will prevent using co on it again (due to lock)
2. edit the file
3. commit the new file + checkout

When using ci to store the new revision, we need to write a small message, try to use something clear and short. The log messages can be seen in the file history, that should help you to know which change has been made and why. The full process is done in the following example.

# co -l fstab
RCS/fstab,v  -->  fstab
revision 1.1 (locked)
done
# echo "something wrong" >> fstab
# ci -u fstab
RCS/fstab,v  <--  fstab
new revision: 1.4; previous revision: 1.3
enter log message, terminated with a single '.' or end of file:
>> I added a mistake on purpose!
>> .
revision 1.4 (unlocked)
done

View changes since last version

Using previous example, we will use rcsdiff to check the changes since the last version.

# co -l fstab
RCS/fstab,v  -->  fstab
revision 1.1 (locked)
done
# echo "something wrong" >> fstab
# rcsdiff -u fstab
--- fstab   2018/10/28 14:28:29 1.1
+++ fstab   2018/10/28 14:30:41
@@ -9,3 +9,4 @@
 52fdd1ce48744600.j /usr/src ffs rw,nodev,nosuid 1 2
 52fdd1ce48744600.e /var ffs rw,nodev,nosuid 1 2
 52fdd1ce48744600.m /data ffs rw,dev,wxallowed,nosuid 1 2
+something wrong

The -u flag is so to produce an unified diff, which I find easier to read. Lines with + shows additions, and lines with - show deletions (there are none in the example).

Use of ci -u

The examples were using ci -u this is because, if you use ci some_file, the file will be saved in the RCS folder but will be missing in its place. You should use co some_file to get it back (in read-only).

# co -l fstab
RCS/fstab,v  -->  fstab
revision 1.1 (locked)
done
# echo "something wrong" >> fstab
# ci -u fstab
RCS/fstab,v  <--  fstab
new revision: 1.4; previous revision: 1.3
enter log message, terminated with a single '.' or end of file:
>> I added a mistake on purpose!
>> .
done
# ls fstab
ls: fstab: No such file or directory
# co fstab
RCS/fstab,v  -->  fstab
revision 1.5
done
# ls fstab
fstab

Using ci -u is very convenient because it prevent the user to forget to checkout the file after commiting the changes.

Show existing revisions of a file

# rlog fstab
RCS file: RCS/fstab,v
Working file: fstab
head: 1.2
branch:
locks: strict
access list:
symbolic names:
keyword substitution: kv
total revisions: 2;     selected revisions: 2
description:
new file
----------------------------
revision 1.2
date: 2018/10/28 14:45:34;  author: solene;  state: Exp;  lines: +1 -0;
Adding a disk
----------------------------
revision 1.1
date: 2018/10/28 14:45:18;  author: solene;  state: Exp;
Initial revision
=============================================================================

We have revisions 1.1 and 1.2, if we want to display the file in its 1.1 revision, we can use the following command:

# co -p1.1 fstab
RCS/fstab,v  -->  standard output
revision 1.1
52fdd1ce48744600.b none swap sw
52fdd1ce48744600.a / ffs rw 1 1
52fdd1ce48744600.l /home ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.d /tmp ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.f /usr ffs rw,nodev 1 2
52fdd1ce48744600.g /usr/X11R6 ffs rw,nodev 1 2
52fdd1ce48744600.h /usr/local ffs rw,wxallowed,nodev 1 2
52fdd1ce48744600.k /usr/obj ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.j /usr/src ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.e /var ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.m /data ffs rw,dev,wxallowed,nosuid 1 2
done

Note that there is no space between the flag and the revision! This is required.

We can see that the command did output some extra informations about the file and “done” at the end of the file. Thoses extra informations are sent to stderr while the actual file content is sent to stdout. That mean if we redirect stdout to a file, we will get the file content.

# co -p1.1 fstab > a_file
RCS/fstab,v  -->  standard output
revision 1.1
done
# cat a_file
52fdd1ce48744600.b none swap sw
52fdd1ce48744600.a / ffs rw 1 1
52fdd1ce48744600.l /home ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.d /tmp ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.f /usr ffs rw,nodev 1 2
52fdd1ce48744600.g /usr/X11R6 ffs rw,nodev 1 2
52fdd1ce48744600.h /usr/local ffs rw,wxallowed,nodev 1 2
52fdd1ce48744600.k /usr/obj ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.j /usr/src ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.e /var ffs rw,nodev,nosuid 1 2
52fdd1ce48744600.m /data ffs rw,dev,wxallowed,nosuid 1 2

Show a diff of a file since a revision

We can use rcsdiff using -r flag to tell it to show the changes between last and one specific revision.

# rcsdiff -u -r1.1 fstab
--- fstab   2018/10/29 14:45:18 1.1
+++ fstab   2018/10/29 14:45:34
@@ -9,3 +9,4 @@
 52fdd1ce48744600.j /usr/src ffs rw,nodev,nosuid 1 2
 52fdd1ce48744600.e /var ffs rw,nodev,nosuid 1 2
 52fdd1ce48744600.m /data ffs rw,dev,wxallowed,nosuid 1 2
+something wrong

With the new OpenSMTPD syntax change which landed with OpenBSD 6.4 release, changes are needed for making opensmtpd to act as a lan relay to a smtp server. This case wasn’t covered in my previous article about opensmtpd, I was only writing about relaying from the local machine, not for a network. Mike (a reader of the blog) shared that it would be nice to have an article about it. Here it is! :)

A simple configuration would look like the following:

listen on em0
listen on lo0

table aliases db:/etc/mail/aliases.db
table secrets db:/etc/mail/secrets.db

action "local" mbox alias <aliases>
action "relay" relay host smtps://myrelay@remote-smtpd.tld auth <secrets>

match for local action "local"
match from local for any action "relay"
match from src 192.168.1.0/24 for action relay

The daemon will listen on em0 interface, and mail delivered from the network will be relayed to remote-smtpd.tld.

For a relay using authentication, the login and passwords must be defined in the file /etc/mail/secrets like this: myrelay login:Pa$$W0rd

smtpd.conf(5) explains creation of /etc/mail/secrets like this:

touch /etc/mail/secrets
chmod 640 /etc/mail/secrets
chown root:_smtpd /etc/mail/secrets

If you are using opensmtpd on a device not always connected on the internet, you may want to see what mail did not go, and force it to be delivered NOW when you are finally connected to the Internet.

We can use smtpctl to show the current queue.

$ doas smtpctl show queue
1de69809e7a84423|local|mta|auth|so@tld|dest@tld|dest@tld|1540362112|1540362112|0|2|pending|406|No MX found for domain

The previous command will report nothing if the queue is empty.

In the previous output, we see that there is one mail from me to dest@tld which is pending due to “NO MX found for domain” (which is normal as I had no internet when I sent the mail).

We need to extract the first field, which is 1de69809e7a84423 in the current example.

In order to tell opensmtpd to deliver it now, we will use the following command:

$ doas smtpctl schedule 1de69809e7a84423
1 envelope scheduled
$ doas smtpctl show queue

My mail was delivered, it’s not in the queue anymore.

If you wish to deliver all enveloppes in the queue, this is as simple as:

$ doas smtpctl schedule all

The default OpenBSD partition layout uses a pre-defined template. If you have a disk more than 356 GB you will have unused space with the default layout (346 GB before 6.4).

It’s possible to create a new partition to use that space if you did not modify the default layout at installation. You only need to start disklabel with flag -E* and type a to add a partition, default will use all remaining space for the partition.

# disklabel -E sd0
Label editor (enter '?' for help at any prompt)
> a
partition: [m]
offset: [741349952]
size: [258863586]
FS type: [4.2BSD]
> w
> q
No label changes.

The new partition here is m. We can format it with:

# newfs /dev/rsd0m

Then, you should add it to your /etc/fstab, for that, use the same uuid as for other partitions, it would look something like 52fdd1ce48744600

52fdd1ce48744600.e /data ffs rw,nodev,nosuid 1 2

It will be auto mounted at boot, you only need to create the folder /data. Now you can do

# mkdir /data
# mount /data

and /data is usable right now.

You can read disklabel(8) and newfs for more informations.

Simple command line to display your installed packages listed by size from smallest to biggest.

$ pkg_info -sa | paste - - - - | sort -n -k 5

Thanks to sthen@ for the command, I was previously using one involving awk which was less readable. paste is often forgotten, it has very specifics uses which can’t be mimic easily with other tools, its purpose is to joins multiples lines into one with some specific rules.

You can easily modify the output to convert the size from bytes to megabytes with awk:

$ pkg_info -sa | paste - - - - | sort -n -k 5 | awk '{ NF=$NF/1024/1024 ; print }'

This divides the last element (using space separator) of each line twice by 1024 and displays the line.

Today I will write about my blog itself. While I started it as my own documentation for some specific things I always forget about (like “How to add a route through a specific interface on FreeBSD”) or to publish my dot files, I enjoyed it and wanted to share about some specific topics.

Then I started the “port of the week” things, but as time goes, I find less of those software and so I don’t have anything to write about. Then, as I run multiples servers, sometimes when I feel that the way I did something is clean and useful, I share it here, as it is a reminder for me I also write it to be helpful for others.

Doing things right is time consuming, but I always want to deliver a polished write. In my opinion, doing things right includes the following:

• explain why something is needed
• explain code examples
• give hints about potential traps
• where to look for official documentation
• provide environment informations like the operating system version used at the writing time
• make the reader to think and get inspired instead of providing a material ready to be copy / pasted brainlessly

I try to keep as much as possible close to those guidelines. I even update from time to time my previous articles to check it still works on the latest operating system version, so the content is still relevant. And until it’s not updated, having the system version let the reader think about “oh, it may have changed” (or not, but it becomes the reader problem).

Now, I want to share about some OpenBSD specifics features, in a way to highlight features. In OpenBSD everything is documented correctly, but as a Human, one can’t read and understand every man page to know what is possible. Here come the highlighting articles, trying to show features, how to use it and where they are documented.

I hope you, reader, like what I write. I am writing here since two years and I still like it.

Following a discussion on the OpenBSD mailing list misc, today I will write about how to manage the priority (as in nice priority) of your daemons or services.

In man page rc(8), one can read:

Before init(8) starts rc, it sets the process priority, umask, and
resource limits according to the “daemon” login class as described in
login.conf(5).  It then starts rc and attempts to execute the sequence of
commands therein.

Using /etc/login.conf we can manage some limits for services and daemon, using their rc script name.

For example, to make jenkins at lowest priority (so it doesn’t make troubles if it builds), using this line will set it to nice 20.

jenkins:priority=20

If you have a file /etc/login.conf.db you have to update it from /etc/login.conf using the software cap_mkdb. This creates a hashed database for faster information retrieval when this file is big. By default, that file doesn’t exist and you don’t have to run cap_mkdb. See login.conf(5) for more informations.

In this article I will show how to configure OpenSMTPD, the default mail server on OpenBSD, to relay mail sent locally to your smtp server. In pratice, this allows to send mail through “localhost” by the right relay, so it makes also possible to send mail even if your computer isn’t connected to the internet. Once connected, opensmtpd will send the mails.

All you need to understand the configuration and write your own one is in the man page smtpd.conf(5). This is only a highlight on was it possible and how to achieve it.

In OpenBSD 6.4 release, the configuration of opensmtpd changed drasticaly, now you have to defines rules and action to do when a mail match the rules, and you have to define those actions.

In the following example, we will see two kinds of relay, the first is through smtp over the Internet, it’s the most likely you will want to setup. And the other one is how to relay to a remote server not allowing relaying from outside.

/etc/mail/smtpd.conf

table aliases file:/etc/mail/aliases
table secrets file:/etc/mail/secrets
listen on lo0

action "local" mbox alias <aliases>
action "relay" relay
action "myserver" relay host smtps://myrelay@perso.pw auth <secrets>
action "openbsd"  relay host localhost:2525

match mail-from "@perso.pw"    for any action "myserver"
match mail-from "@openbsd.org" for any action "openbsd"
match for local action "local"
match for any action "relay"

I defined 2 actions, one from “myserver”, it has a label “myrelay” and we use auth <secrets> to tell opensmtpd it needs authentication.

The other action is “openbsd”, it will only relay to localhost on port 2525.

To use them, I define 2 matching rules of the very same kind. If the mail that I want to send match the @domain-name, then choose relay “myserver” or “openbsd”.

The “openbsd” relay is only available when I create a SSH tunnel, binding the local port 25 of the remote server to my port 2525, with flags -L 2525:127.0.0.1:25.

For a relay using authentication, the login and passwords must be defined in the file /etc/mail/secrets like this: myrelay login:Pa$$W0rd

smtpd.conf(5) explains creation of /etc/mail/secrets like this:

touch /etc/mail/secrets
chmod 640 /etc/mail/secrets
chown root:_smtpd /etc/mail/secrets

Now, restarts your server. Then if you need to send mails, just use “mail” command or localhost as a smtp server. Depending on your From address, a different relay will be used.

Deliveries can be checked in /var/log/maillog log file.

See mails in queue

doas smtpctl show queue

Try to deliver now

doas smtpctl schedule all

Today I will cover a specific topic on OpenBSD networking. If you are using a laptop, you may switch from ethernet to wireless network from time to time. There is a simple way to keep the network instead of having to disconnect / reconnect everytime.

It’s possible to aggregate your wireless and ethernet devices into one trunk pseudo device in failover mode, which give ethernet the priority if connected.

To achieve this, it’s quite simple. If you have devices em0 and iwm0 create the following files.

/etc/hostname.em0

up

/etc/hostname.iwm0

join "office_network"  wpakey "mypassword"
join "my_home_network" wpakey "9charshere"
join "roaming phone"   wpakey "something"
join "Public Wifi"
up

/etc/hostname.trunk0

trunkproto failover trunkport em0 trunkport iwm0
dhcp

As you can see in the wireless device configuration we can specify multiples network to join, it is a new feature that will be available from 6.4 release.

You can enable the new configuration by running sh /etc/netstart as root.

This setup is explained in trunk(4) man page and in the OpenBSD FAQ as well.