Backup Solution - rsnapshot¶
- Know how to install additional repositories and snapshots from the command-line
- Know about mounting filesystems external of your machine (external hard drive, remote filesystem, etc.)
- Know how to use an editor (
viis used here, but you can use your favorite editor)
- Know a little BASH scripting
- Know how to modify the crontab for the root user
- Knowledge of SSH public and private keys (only if you plan to run remote backups from another server)
rsnapshot is a very powerful backup utility that can be installed on any Linux-based machine. It can either back up a machine locally, or you can back up multiple machines, say servers for instance, from a single machine.
rsync and is written entirely in perl with no library dependencies, so there are no weird requirements for installation. In the case of Rocky Linux, you should generally be able to install rnapshot by using the EPEL repository. After the initial release of Rocky Linux 9.0, there was a period of time when the EPEL did not contain the rsnapshot package. That has since been rectified, but we have included a method of installing from source just in case this should happen again.
This documentation covers the installation of rsnapshot on Rocky Linux only.
All commands shown here are called from the command-line on your server or workstation unless otherwise noted.
Installing The EPEL repository¶
We need the EPEL software repository from Fedora to install rsnapshot. To install the repository, just use this command:
sudo dnf install epel-release
The repository should now be active.
Install the rsnapshot Package¶
Next, install rsnapshot and some other needed tools, which are probably already installed:
sudo dnf install rsnapshot openssh-server rsync
If there are any missing dependencies, those will show up and you simply need to answer the prompt to continue. For example:
dnf install rsnapshot Last metadata expiration check: 0:00:16 ago on Mon Feb 22 00:12:45 2021. Dependencies resolved. ======================================================================================================================================== Package Architecture Version Repository Size ======================================================================================================================================== Installing: rsnapshot noarch 1.4.3-1.el8 epel 121 k Installing dependencies: perl-Lchown x86_64 1.01-14.el8 epel 18 k rsync x86_64 3.1.3-9.el8 baseos 404 k Transaction Summary ======================================================================================================================================== Install 3 Packages Total download size: 543 k Installed size: 1.2 M Is this ok [y/N]: y
Installing rsnapshot From Source¶
Installing rsnapshot from source is not difficult. It does come with a downside, however, in that if there is a new version released, it would require a fresh install from source to update the version, whereas the EPEL installation method would keep you up-to-date with a simple
Installing Development Tools and Downloading the Source¶
As stated, the first step here is to install the 'Development Tools' group:
dnf groupinstall 'Development Tools'
You also need a few other packages:
dnf install wget unzip rsync openssh-server
Next we will need to download the source files from the GitHub repository. You can do this multiple ways, but the easiest in this case is probably just to download the ZIP file from the repository.
- Go to https://github.com/rsnapshot/rsnapshot
- Click on the Green "Code" button on the right
- Right-click on the "Download ZIP" and copy the link location
curlto download the copied link. Example:
- Unzip the
Building the Source¶
Now that we've got everything on our machine, the next step is to build. When we unzipped the
master.zip file, we ended up with an
rsnapshot-master directory. We will need to change into this for our build procedure. Note that our build is using all of the package defaults, so if you want something else, you'll need to do a little investigation. Also, these steps are directly taken from the GitHub Installation page:
authogen.sh script to generate the configure script:
You may get several lines that look like this:
fatal: not a git repository (or any of the parent directories): .git
These are not fatal.
Next, we need to run
configure with the configuration directory set:
sudo make install
During all of this, the
rsnapshot.conf file will be created as
rsnapshot.conf.default. We need to copy this over to
rsnapshot.conf and then edit it to fit what we need on our system.
sudo cp /etc/rsnapshot.conf.default /etc/rsnapshot.conf
This covers copying the configuration file over. The section below on "Configuring rsnapshot" will cover the changes needed in this configuration file.
Mounting A Drive or Filesystem For Backup¶
In this step, we show how to mount a hard drive, such as an external USB hard drive, that will be used to back up your system. This particular step is only necessary if you are backing up a single machine or server, as seen in our first example below.
- Plug in the USB drive.
dmesg | grep sdwhich should show you the drive you want to use. In this case, it'll be called sda1.
EXT4-fs (sda1): mounting ext2 file system using the ext4 subsystem.
- Unfortunately (or fortunately depending on your opinion) most modern Linux desktop operating systems automount the drive if they can. This means that, depending on various factors, rsnapshot might lose track of the hard drive. We want the drive to "mount" or make its files available in the same place every time.
To do that, take the drive information revealed in the
dmesgcommand above and type
mount | grep sda1, which should show something like this:
/dev/sda1 on /media/username/8ea89e5e-9291-45c1-961d-99c346a2628a
sudo umount /dev/sda1to unmount your external hard drive.
- Next, create a new mount point for the backup:
sudo mkdir /mnt/backup
- Now mount the drive to your backup folder location:
sudo mount /dev/sda1 /mnt/backup
- Now type
mount | grep sda1again, and you should see something like this:
/dev/sda1 on /mnt/backup type ext2 (rw,relatime)
- Next create a directory that must exist for the backup to continue on the mounted drive. We are using a folder called "storage" for this example:
sudo mkdir /mnt/backup/storage
Note that for a single machine, you will have to either repeat the
mount steps each time the drive is plugged in again, or each time the system reboots, or automate these commands with a script.
We recommend automation. Automation is the sysadmin way.
This is the most important step. It's easy to make a mistake when making changes to the configuration file. The rsnapshot configuration requires tabs for any separation between elements, and a warning to that effect is at the very top of the configuration file.
A space character will cause the entire configuration—and your backup—to fail. For instance, near the top of the configuration file is a section for the
# SNAPSHOT ROOT DIRECTORY #. If you were adding this in from scratch, you would type
snapshot_root then TAB and then type
The best thing is that the default configuration that comes with rsnapshot only needs minor changes to make it work for a backup of a local machine. It's always a good idea, though, to make a backup copy of the configuration file before you start editing:
cp /etc/rsnapshot.conf /etc/rsnapshot.conf.bak
Basic Machine or Single Server Backup¶
In this case, rsnapshot is going to be run locally to back up a particular machine. In this example, we'll break down the configuration file, and show you exactly what you need to change.
You will need to use
vi (or edit with your favorite editor) to open the /etc/rsnapshot.conf file.
The first thing to change is the snapshot_root setting which by default has this value:
We need to change this to our mount point that we created above plus the addition of "storage".
We also want to tell the backup NOT to run if the drive is not mounted. To do this, remove the "#" sign (also called a remark, pound sign, number sign, hash symbol, etc.) next to
no_create_root so that it looks like this:
Next go down to the section titled
# EXTERNAL PROGRAM DEPENDENCIES # and remove the comment (again, the "#" sign) from this line:
So that it now reads:
While we do not need
cmd_ssh for this particular configuration, we will need it for our other option below and it doesn't hurt to have it enabled. So find the line that says:
And remove the "#" sign so that it looks like this:
Next we need to skip down to the section titled
# BACKUP LEVELS / INTERVALS #
This has been changed from earlier versions of rsnapshot from
hourly, daily, monthly, yearly to
alpha, beta, gamma, delta. Which is a bit confusing. What you need to do is add a remark to any interval that you won't be using. In the configuration, delta is already remarked out.
For this example, we aren't going to be running any other increments other than a nightly backup, so just add a remark to alpha and gamma so that the configuration looks like this when you are done:
#retain alpha 6 retain beta 7 #retain gamma 4 #retain delta 3
Now skip down to the logfile line, which by default should read:
And remove the remark so that it is enabled:
Finally, skip down to the
### BACKUP POINTS / SCRIPTS ### section and add any directories that you want to add in the
# LOCALHOST section, remember to use TAB rather than SPACE between elements!
For now write your changes (
SHIFT :wq! for
vi) and exit the configuration file.
Checking The Configuration¶
We want to make sure that we didn't add spaces or any other glaring errors to our configuration file while we were editing it. To do this, we run rsnapshot against our configuration with the
rsnapshot configtest will show
Syntax OK if there are no errors in the configuration.
You should get into the habit of running
configtest against a particular configuration. The reason for that will be more evident when we get into the Multiple Machine or Multiple Server Backups section.
configtest against a particular configuration file, run it with the -c option to specify the configuration:
rsnapshot -c /etc/rsnapshot.conf configtest
Running The Backup The First Time¶
Everything has checked out, so it's time to go ahead and run the backup for the first time. You can run this in test mode first if you like, so that you can see what the backup script is going to do.
Again, to do this you don't necessarily have to specify the configuration in this case, but you should get into the habit of doing so:
rsnapshot -c /etc/rsnapshot.conf -t beta
Which should return something like this, showing you what will happen when the backup is actually run:
echo 1441 > /var/run/rsnapshot.pid mkdir -m 0755 -p /mnt/backup/storage/beta.0/ /usr/bin/rsync -a --delete --numeric-ids --relative --delete-excluded \ /home/ /mnt/backup/storage/beta.0/localhost/ mkdir -m 0755 -p /mnt/backup/storage/beta.0/ /usr/bin/rsync -a --delete --numeric-ids --relative --delete-excluded /etc/ \ /mnt/backup/storage/beta.0/localhost/ mkdir -m 0755 -p /mnt/backup/storage/beta.0/ /usr/bin/rsync -a --delete --numeric-ids --relative --delete-excluded \ /usr/local/ /mnt/backup/storage/beta.0/localhost/ touch /mnt/backup/storage/beta.0/
Once you are satisfied with the test, go ahead and run it manually the first time without the test:
rsnapshot -c /etc/rsnapshot.conf beta
When the backup finishes, navigate to /mnt/backup and take a look at the directory structure that was created there. There will be a
storage/beta.0/localhost directory, followed by the directories that you specified to backup.
Each time the backup is run, it will create a new beta increment, 0-6, or 7 days worth of backups. The newest backup will always be beta.0 whereas yesterday's backup will always be beta.1.
The size of each of these backups will appear to take up the same amount (or more) of disk space, but this is because of rsnapshot's use of hard links. To restore files from yesterday's backup, you would simply copy them back from beta.1's directory structure.
Each backup is only an incremental backup from the previous run, BUT, because of the use of hard links, each backup directory, contains either the file or the hard-link to the file in whichever directory it was actually backed up in.
So to restore files, you don't have to pick and choose which directory or increment to restore them from, just what time stamp the backup should have that you are restoring. It's a great system and uses far less disk space than many other backup solutions.
Setting The Backup To Run Automatically¶
Once everything has been tested and we know that things will work without issue, the next step is to set up the crontab for the root user, so that all of this can be done automatically every day:
sudo crontab -e
If you haven't run this before, choose vim.basic as your editor or your own editor preference when the
Select an editor line comes up.
We are going to set our backup to automatically run at 11 PM, so we will add this to the crontab:
## Running the backup at 11 PM 00 23 * * * /usr/bin/rsnapshot -c /etc/rsnapshot.conf beta`
Multiple Machine or Multiple Server Backups¶
Doing backups of multiple machines from a machine with a RAID array or large storage capacity, either on premises or from across the Internet works very well.
If running these backups from across the Internet, you need to make sure that both locations have adequate bandwidth for the backups to occur. You can use rsnapshot to synchronize an on-site server with an off-site backup array or backup server to improve data redundancy.
We are assuming that you are running rsnapshot from a machine remotely, on-premise. This exact configuration can be duplicated, as indicated above, remotely off-premise as well.
In this case, you will want to install rsnapshot on the machine that is doing all of the backups. We are also assuming:
- That the servers you will be backing up to, have a firewall rule that allows the remote machine to SSH into it
- That each server that you will be backing up has a recent version of
rsyncinstalled. For Rocky Linux servers, run
dnf install rsyncto update your system's version of
- That you've either connected to the machine as the root user, or that you have run
sudo -sto switch to the root user.
SSH Public / Private Keys¶
For the server that will be running the backups, we need to generate an SSH key-pair for use during the backups. For our example, we will be creating RSA keys.
If you already have a set of keys generated, you can skip this step. You can find out by doing an
ls -al .ssh and looking for an
id_rsa.pub key pair. If none exists, use the following link to set up keys for your machine and the server(s) that you want to access:
The configuration file needs to be just like the one we created for the Basic Machine or Single Server Backup above, except that we want to change some of the options.
The snapshot root can be reverted back to the default like so:
And this line:
... can be commented out again:
The other difference here is that each machine will have its very own configuration. Once you get used to this, you'll simply copy one of your existing configuration files over to a new name and then modify it to fit any additional machines that you want to backup.
For now, we want to modify the configuration file just like we did above, and then save it. Then copy that file as a template for our first server:
cp /etc/rsnapshot.conf /etc/rsnapshot_web.conf
We want to modify the new configuration file and create the log and lockfile with the machine's name:
Next, we want to modify rsnapshot_web.conf so that it includes the directories we want to back up. The only thing that is different here is the target.
Here's an example of the web.ourdomain.com configuration:
### BACKUP POINTS / SCRIPTS ### backup email@example.com:/etc/ web.ourourdomain.com/ backup firstname.lastname@example.org:/var/www/ web.ourourdomain.com/ backup email@example.com:/usr/local/ web.ourdomain.com/ backup firstname.lastname@example.org:/home/ web.ourdomain.com/ backup email@example.com:/root/ web.ourdomain.com/
Checking The Configuration and Running The Initial Backup¶
Just like before, we can now check the configuration to make sure it is syntactically correct:
rsnapshot -c /etc/rsnapshot_web.conf configtest
And just like before, we are looking for the
Syntax OK message. If all is well, we can execute the backup manually:
/usr/bin/rsnapshot -c /etc/rsnapshot_web.conf beta
Assuming that everything works alright, we can then create the configuration files for the mail server (rsnapshot_mail.conf) and portal server (rsnapshot_portal.conf), test them, and do a trial backup.
Automating The Backup¶
Automating backups for the multiple machine/server version is slightly different. We want to create a bash script to call the backups in order. When one finishes the next will start. This script will look something like this and be stored in /usr/local/sbin:
With the content:
#!/bin/bash/ # script to run rsnapshot backups in succession /usr/bin/rsnapshot -c /etc/rsnapshot_web.conf beta /usr/bin/rsnapshot -c /etc/rsnapshot_mail.conf beta /usr/bin/rsnapshot -c /etc/rsnapshot_portal.conf beta
chmod +x /usr/local/sbin/backup_all
Then create the crontab for root to run the backup script:
And add this line:
## Running the backup at 11 PM 00 23 * * * /usr/local/sbin/backup_all
Reporting The Backup Status¶
To make sure that everything is backing up according to plan, you might want to send the backup log files to your email. If your are running multiple machine backups using rsnapshot, each log file will have its own name, which you can then send to your email for review by Using the postfix For Server Process Reporting procedure.
Restoring a Backup¶
Restoring a backup, either a few files or a complete restore, involves copying the files you want from the directory with the date that you want to restore from back to your machine. Simple!
Conclusions and Other Resources¶
Getting the setup right with rsnapshot is a little daunting at first, but can save you loads of time backing up your machines or servers.
rsnapshot is very powerful, very fast, and very economical on disk space usage. You can find more information on rsnapshot, by visiting rsnapshot.org
Author: Steven Spencer
Contributors: Ezequiel Bruni