About me: My name is Solène Rapenne, pronouns she/her. I like learning and sharing knowledge. Hobbies: '(BSD OpenBSD Qubes OS Lisp cmdline gaming security QubesOS internet-stuff). I love percent and lambda characters. OpenBSD developer solene@. No AI is involved in this blog.

Contact me: solene at dataswamp dot org or @solene@bsd.network (mastodon).

You can sponsor my work financially if you want to help me writing this blog and contributing to Free Software as my daily job.

Configuration deployment made easy with drist

Written by Solène, on 29 November 2018.
Tags: #unix #drist #automation

Comments on Fediverse/Mastodon

Hello, in this article I will present you my deployement tool drist (if you speak Russian, I am already aware of what you think). It reached a feature complete status today and now I can write about it.

As a system administrator, I started using salt a few years ago. And honestly, I can not cope with it anymore. It is slow, it can get very complicated for some tasks like correctly ordering commands and a configuration file can become a nightmare when you start using condition in it.

History

I also tried alternatives like ansible, puppet, Rex etc… One day, when lurking in the ports tree, I found sysutils/radmind which got a lot interest from me even if it is really poorly documented. It is a project from 1995 if I remember correctly, but I liked the base idea. Radmind works with files, you create a known working set of files for your system, and you can propagate that whole set to other machines, or see differences between the reference and the current system. Sets could be negative, meaning that the listed files should not be present on the system, but it was also possible to add extra sets for specific hosts. The whole thing is really really cumbersome, this requires a lot of work, I found little documentation etc… so I did not used it but, that lead me to write my own deployment tool using ideas from radmind (working with files) and from Rex (using a script for doing changes).

Concept

drist aims at being simple to understand and pluggable with standard tools. There is no special syntax to learn, no daemon to run, no agent, and it relies on base tools like awk, sed, ssh and rsync.

drist is cross platform as it has a few requirements but it is not well suited for deploying on too much differents operating systems.

When executed, drist will execute six steps in a specific order, you can use only steps you need.

Shamelessly copied from the man page, explanations after:

  1. If folder files exists, its content is copied to server rsync(1).
  2. If folder files-HOSTNAME exists, its content is copied to server using rsync(1).
  3. If folder absent exists, filenames in it are deleted on server.
  4. If folder absent-HOSTNAME exists, filenames in it are deleted on server.
  5. If file script exists, it is copied to server and executed there.
  6. If file script-HOSTNAME exists, it is copied to server and executed there.

In the previous list, all the existences checks are done from the current working directory where drist is started. The text HOSTNAME is replaced by the output of uname -n of the remote server, and files are copied starting from the root directory.

drist does not do anything more. In a more litteral manner, it copies files to the remote server, using a local filesystem tree (folder files). It will delete on the remote server all files present in the local filesystem tree (folder absent), and it will run on the remote server a script named script.

Each of theses can be customized per-host by adding a “-HOSTNAME” suffix to the folder or file name, because experience taught me that some hosts does require specific configuration.

If a folder or a file does not exist, drist will skip it. So it is possible to only copy files, or only execute a script, or delete files and execute a script after.

Drist usage

The usage is pretty simple. drist has 3 flags which are optionals.

  • -n flag will show what happens (simuation mode)
  • -s flag tells drist to use sudo on the remote host
  • -e flag with a parameter will tell drist to use a specific path for the sudo program

The remote server address (ssh format like user@host) is mandatory.

$ drist my_user@my_remote_host

drist will look at files and folders in the current directory when executed, this allow to organize as you want using your filesystem and a revision control system.

Simple examples

Here are two examples to illustrate its usage. The examples are easy, for learning purpose.

Deploying ssh keys

I want to easily copy my users ssh keys to a remote server.

$ mkdir drist_deploy_ssh_keys
$ cd drist_deploy_ssh_keys
$ mkdir -p files/home/my_user1/.ssh
$ mkdir -p files/home/my_user2/.ssh
$ cp -fr /path/to/key1/id_rsa files/home/my_user1/.ssh/
$ cp -fr /path/to/key2/id_rsa files/home/my_user2/.ssh/
$ drist user@remote-host
Copying files from folder "files":
	/home/my_user1/.ssh/id_rsa
	/home/my_user2/.ssh/id_rsa

Deploying authorized_keys file

We can easily create the authorized_key file by using cat.

$ mkdir drist_deploy_ssh_authorized
$ cd drist_deploy_ssh_authorized
$ mkdir -p files/home/user/.ssh/
$ cat /path/to/user/keys/*.pub > files/home/user/.ssh/authorized_keys
$ drist user@remote-host
Copying files from folder "files":
	/home/user/.ssh/authorized_keys

This can be automated using a makefile running the cat command and then running drist.

all:
	cat /path/to/keys/*.pub > files/home/user.ssh/authorized_keys
drist user@remote-host

Installing nginx on FreeBSD

This module (aka a folder which contain material for drist) will install nginx on FreeBSD and start it.

$ mkdir deploy_nginx
$ cd deploy_nginx
$ cat >script <<EOF
#!/bin/sh
test -f /usr/local/bin/nginx
if [ $? -ne 0 ]; then
	pkg install -y nginx
fi
sysrc nginx_enable=yes
service nginx restart
EOF
$ drist user@remote-host
Executing file "script":
	Updating FreeBSD repository catalogue...
	FreeBSD repository is up to date.
	All repositories are up to date.
	The following 1 package(s) will be affected (of 0 checked):
	
	New packages to be INSTALLED:
	        nginx: 1.14.1,2
	
	Number of packages to be installed: 1
	
	The process will require 1 MiB more space.
	421 KiB to be downloaded.
	[1/1] Fetching nginx-1.14.1,2.txz: 100%  421 KiB 430.7kB/s    00:01
	Checking integrity... done (0 conflicting)
	[1/1] Installing nginx-1.14.1,2...
	===> Creating groups.
	Using existing group 'www'.
	===> Creating users
	Using existing user 'www'.
	[1/1] Extracting nginx-1.14.1,2: 100%
	Message from nginx-1.14.1,2:
	
	===================================================================
	Recent version of the NGINX introduces dynamic modules support.  In
	FreeBSD ports tree this feature was enabled by default with the DSO
	knob.  Several vendor's and third-party modules have been converted
	to dynamic modules.  Unset the DSO knob builds an NGINX without
	dynamic modules support.
	
	To load a module at runtime, include the new `load_module'
	directive in the main context, specifying the path to the shared
	object file for the module, enclosed in quotation marks.  When you
	reload the configuration or restart NGINX, the module is loaded in.
	It is possible to specify a path relative to the source directory,
	or a full path, please see
	https://www.nginx.com/blog/dynamic-modules-nginx-1-9-11/ and
	http://nginx.org/en/docs/ngx_core_module.html#load_module for
	details.
	
	Default path for the NGINX dynamic modules is
	
	/usr/local/libexec/nginx.
	===================================================================
	nginx_enable:  -> yes
	Performing sanity check on nginx configuration:
	nginx: the configuration file /usr/local/etc/nginx/nginx.conf syntax is ok
	nginx: configuration file /usr/local/etc/nginx/nginx.conf test is successful
	nginx not running? (check /var/run/nginx.pid).
	Performing sanity check on nginx configuration:
	nginx: the configuration file /usr/local/etc/nginx/nginx.conf syntax is ok
	nginx: configuration file /usr/local/etc/nginx/nginx.conf test is successful
	Starting nginx.

More complex example

Now I will show more complexes examples, with host specific steps. I will not display the output because the previous output were sufficient enough to give a rough idea of what drist does.

Removing someone ssh access

We will reuse an existing module here, a user should not be able to login anymore on its account on the servers using the ssh key.

$ cd ssh
$ mkdir -p absent/home/user/.ssh/
$ touch absent/home/user/.ssh/authorized_keys
$ drist user@server

Installing php on FreeBSD

The following module will install php and remove the opcache.ini file, and will install php72-pdo_pgsql if it is run on server production.domain.private.

$ mkdir deploy_php && cd deploy_php
$ mkdir -p files/usr/local/etc
$ cp /some/correct/config.ini files/usr/local/etc/php.ini
$ cat > script <<EOF
#!/bin/sh
test -f /usr/local/etc/php-fpm.conf || pkg install -f php-extensions
sysrc php_fpm_enable=yes
service php-fpm restart
test -f /usr/local/etc/php/opcache.ini || rm /usr/local/etc/php/opcache.ini
EOF
$ cat > script-production.domain.private <<EOF
#!/bin/sh
test -f /usr/local/etc/php/pdo_pgsql.ini || pkg install -f php72-pdo_pgsql
service php-fpm restart
EOF

The monitoring machine

This one is unique and I would like to avoid applying its configuration against another server (that happened to me once with salt and it was really really bad). So I will just do all the job using the hostname specific cases.

$ mkdir my_unique_machine && cd my_unique_machine
$ mkdir -p files-unique-machine.private/usr/local/etc/{smokeping,munin}
$ cp /good/config files-unique-machine.private/usr/local/etc/smokeping/config
$ cp /correct/conf files-unique-machine.private/usr/local/etc/munin/munin.conf
$ cat > script-unique-machine.private <<EOF
#!/bin/sh
pkg install -y smokeping munin-master munin-node
munin-configure --shell --suggest | sh
sysrc munin_node_enable=yes
sysrc smokeping_enable=yes
service munin-node restart
service smokeping restart
EOF
$ drist user@incorrect-host
$ drist user@unique-machine.private
Copying files from folder "files-unique-machine.private":
	/usr/local/etc/smokeping/config
	/usr/local/etc/munin/munin.conf
Executing file "script-unique-machine.private":
    [...]

Nothing happened on the wrong system.

Be creative

Everything can be automated easily. I have some makefile in a lot of my drist modules, because I just need to type “make” to run it correctly. Sometimes it requires concatenating files before being run, sometimes I do not want to make mistake or having to remember on which module apply on which server (if it’s specific), so the makefile does the job for me.

One of my drist module will look at all my SSL certificates from another module, and make a reed-alert configuration file using awk and deploying it on the monitoring server. All I do is typing “make” and enjoy my free time.

How to get it and install it

  • Drist can be downloaded at this address.
  • Sources can be cloned using git clone git://bitreich.org/drist

In the sources folder, type “make install” as root, that will copy drist binary to /usr/bin/drist and its man page to /usr/share/man/man1/drist.1

For copying files, drist requires rsync on both local and remote hosts.

For running the script file, a sh compatible shell is required (csh is not working).