How to run scripts on start up?
How can I run scripts automatically when Ubuntu starts up so I don’t have to run them manually after startup?
If someone could also show both WHEN and WHERE that would be awesome. I say this because I know there are at least 2 ways to start a script that will fire before other applications have been started (like X11)
This entire answer thread is a mess. The Stack Exchange format doesn’t seem to be best suited for this question
+1 to @GabrielFair. A LARGE part of the problem is that the original question and answer are TEN YEARS OLD. I’d also add that there are too many ways to solve this problem. What happened to the simplicity of the Unix philosophy?! Request someone knowledgeable and with sufficient points rewrite this post, or add a new, up-to-date, definitive answer for modern os versions.
10 Answers 10
One approach is to add an @reboot cron task:
- Running crontab -e will allow you to edit your cron.
- Adding a line like this to it:
This is awesome. So far this seems better than rc.local since the system seems more setup by this point (PATH, etc). It is odd that it is so hard to call something after system startup..
Depending on what sort of scripts you need to run.. For services and the like you should use upstart. But for a user script these should be launched as session scripts by gnome! Have a look under System > Preferences > Startup Applications.
On a side note if you need some scripts to be run on terminal login you can add them to the .bash_login file in your home directory.
For 14.04 and older
A simple command (one which doesn’t need to remain running) could use an Upstart job like:
start on startup task exec /path/to/command Save this in a .conf file in /etc/init (if you need it to run as root when the system boots up), or in ~/.config/upstart (if you need it to run as your user when you log in).
Considering how SO and StackExchange runs, could you give an example of an upstart script and where it would be placed? That would make this a much better answer. Your link says it’s not being maintained and to look at the upstart cookbook, which is huuuge. I don’t have too much of an idea where to start.
Please update this answer to explain what to do on systems running systemd rather than upstart (Ubuntu 15.04+).
This answer does not make sense to me. The applications listed in system->pref->startup applications can neither be found in /etc/init/ nor in ~/.config/upstart . So where are startup applications defined?
You can add commands to /etc/rc.local :
This executes the commands as root.
To execute commands as a specific user, use sudo -i -u ( -i to also run the login shell). For example, to establish a persistent SSH tunnel, where myhost is definde in johndoe s ~/.ssh/config file:
sudo -i -u johndoe autossh -nNT -L 1234:localhost:1234 myhost Note that if /etc/rc.local did not exist (as is the case on Ubuntu since 16.04), you need to add a shebang line at the top (e.g. #!/bin/bash ), and ensure the file is executable:
This most directly answers the question: how to simply execute some scripts when your system boots. upstart does a more complex task: starts daemon processes.
Note that if you make this file yourself (like I did), then you will have to change file to executable with chmod 755 rc.local , and add #!/bin/bash to the first line.
For 15.04 and later:
To run a (short-lived) 1 command at startup using systemd , you can use a systemd unit of type OneShot . For example, create /etc/systemd/system/foo.service containing:
[Unit] Description=Job that runs your user script [Service] ExecStart=/some/command Type=oneshot RemainAfterExit=yes [Install] WantedBy=multi-user.target sudo systemctl daemon-reload sudo systemctl enable foo.service Essentially, this is just converting a typical Upstart job to a systemd one (see Systemd for Upstart users).
You can run multiple commands from the same service file, using multiple ExecStart lines:
[Service] ExecStart=/some/command ExecStart=/another/command some args ExecStart=-/a/third/command ignore failure The command must always be given with the full path. If any command fails, the rest aren’t run. A — before the path tells systemd to ignore a non-zero exit status (instead of considering it a failure).
For user sessions, you can create the systemd unit in ~/.config/systemd/user instead. This should work with 16.04 onwards, but not earlier releases of Ubuntu with systemd (since those still used Upstart for user sessions). User session units can be controlled with the same commands as with system services, but with the —user option added:
systemctl --user daemon-reload systemctl --user status foo.service Shell syntax
Note that, unlike Upstart, systemd doesn’t run the Exec* commands through a shell. It performs some limited variable expansion and multiple command (separated by ; ) itself, but that’s about it as far as shell-like syntax goes. For anything more complicated, say redirection or pipes, wrap your command in sh -c ‘. ‘ or bash -c ‘. ‘ .
1 As opposed to long-lived daemons.
is it possible to set a priority on the job? or specify that it depends on another service to be started first?
@r3wt yes, there are different ways to do that. The WantedBy used here, for example, makes it start when the multi-user.target is reached. You can use Before , After , Requires , etc. See man systemd.unit
You’re welcome. — Btw, the RemainAfterExit depends on the service you start and its desired behaviour. For instance, /bin/df -h would should have RemainAfterExit=no .
The difference comes when you issue systemctl status foo : with RemainAfterExit=yes it says active while with RemainAfterExit=no it doesn’t. However.
There are different ways to automatically run commands:
- The upstart system will execute all scripts from which it finds a configuration in directory /etc/init . These scripts will run during system startup (or in response to certain events, e.g., a shutdown request) and so are the place to run commands that do not interact with the user; all servers are started using this mechanism. You can find a readable introduction to at: http://upstart.ubuntu.com/getting-started.html the man pages man 5 init and man 8 init give you the full details.
- A shell script named .gnomerc in your home directory is automatically sourced each time you log in to a GNOME session. You can put arbitrary commands in there; environment variables that you set in this script will be seen by any program that you run in your session. Note that the session does not start until the .gnomerc script is finished; therefore, if you want to autostart some long-running program, you need to append & to the program invocation, in order to detach it from the running shell.
- The menu option System -> Preferences -> Startup Applications allows you to define what applications should be started when your graphical session starts (Ubuntu predefines quite some), and add or remove them to your taste. This has almost the same purpose and scope of the .gnomerc script, except you don’t need to know sh syntax (but neither can you use any sh programming construct).
3) «This has almost the same purpose and scope of the .gnomerc script», except .gnomerc apparently runs before loading Unity, and Startup Applications apparently runs after loading Unity. I had to run a program that sits on Unity’s menu bar and it made a huge difference in this case!
sudo update-rc.d myscript.sh defaults , where /etc/init.d/myscript.sh is your script, also runs it at startup.
$HOME/.config/autostart contains the startup application list. .desktop files in this folder will be executed on startup. It may need executable permission ( chmod +x startup.desktop ).
Sample example for .desktop file:
[Desktop Entry] Type=Application Exec="" Hidden=false NoDisplay=false X-GNOME-Autostart-enabled=true Name=Startup Script Here «» is replaced with path to your script.sh If you place your script myscript in /usr/local/bin so that it can be executed directly by command, you can write myscript instead of «» .
Sample example of myscript.sh :
Result: .desktop file will be launched from $HOME/.config/autostart which execute script by Exec=
Doesn’t work on Ubuntu 18.04. $HOME/.config/autostart contains but one file for dropbox, which BTW isn’t executable.
Worked for me on Ubuntu 18.04.1. Btw. I added the desktop entry via Search -> Startup applications. The outcome is exactly the same though.
or you can hardcode your password in your script , like below Exec=gnome-terminal — sh -c ‘echo «YOURPASSWORD» | sudo -S sh PATH_TO_YOUR_SCRIPT && sleep 1 && printf «\n»‘
For simple things you can add a command in System->Preferences->Sessions pointing to the location of your script.
Alternatively you can add it to /etc/init.d/rc.local or make an upstart job if it’s a more low level stuff.
cron answer implemented different from top voted
This answer still uses cron but uses a different method than the top voted answer. This works since Ubuntu 16.04 but probably supported much sooner. It’s just that I started using cron to run jobs when computer boots up since 16.04.
When does cron run?
In comments someone asked «when do they run?». You can tell in syslog / journalctl:
$ journalctl -b | grep cron Jan 02 16:54:40 alien cron[919]: (CRON) INFO (pidfile fd = 3) Jan 02 16:54:40 alien cron[919]: (CRON) INFO (Running @reboot jobs) Jan 02 16:54:40 alien systemd[1]: Started Run anacron jobs. Jan 02 16:54:40 alien anacron[949]: Anacron 2.3 started on 2018-01-02 Jan 02 16:54:40 alien anacron[949]: Normal exit (0 jobs run) Jan 02 16:54:40 alien CRON[952]: pam_unix(cron:session): session opened for user root by (uid=0) Jan 02 16:54:40 alien CRON[954]: pam_unix(cron:session): session opened for user root by (uid=0) Jan 02 16:54:40 alien CRON[951]: pam_unix(cron:session): session opened for user root by (uid=0) Jan 02 16:54:40 alien CRON[950]: pam_unix(cron:session): session opened for user root by (uid=0) Jan 02 16:54:40 alien CRON[985]: (root) CMD ( /usr/local/bin/cron-reboot-cycle-grub-background) Jan 02 16:54:40 alien CRON[954]: pam_unix(cron:session): session closed for user root Jan 02 16:54:40 alien cron[919]: sendmail: Cannot open smtp.gmail.com:587 Jan 02 16:54:40 alien CRON[952]: pam_unix(cron:session): session closed for user root Jan 02 16:54:40 alien cron[919]: sendmail: Cannot open smtp.gmail.com:587 Jan 02 16:54:40 alien CRON[950]: pam_unix(cron:session): session closed for user root One thing to note is cron can email you status of jobs run and @reboot jobs run so early network manager and email won’t be running unless you put a sleep command into your script(s).
Where to put your scripts
Put your scripts in the directory /etc/cron.d :
$ ll /etc/cron.d total 44 drwxr-xr-x 2 root root 4096 Nov 26 19:53 ./ drwxr-xr-x 139 root root 12288 Dec 31 13:58 ../ -rw-r--r-- 1 root root 244 Dec 28 2014 anacron -rw-r--r-- 1 root root 148 Feb 18 2017 cycle-grub-background -rw-r--r-- 1 root root 138 Mar 5 2017 display-auto-brightness -rw-r--r-- 1 root root 460 Nov 26 19:53 nvidia-hdmi-sound -rw-r--r-- 1 root root 102 Feb 9 2013 .placeholder -rw-r--r-- 1 root root 224 Nov 19 2016 touch-vmlinuz -rw-r--r-- 1 root root 700 Aug 5 11:15 turn-off-hyper-threading What does a script look like?
Here are a couple of scripts I have setup to run each boot:
$ cat /etc/cron.d/cycle-grub-background SHELL=/bin/sh PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin @reboot root /usr/local/bin/cron-reboot-cycle-grub-background $ cat /etc/cron.d/touch-vmlinuz SHELL=/bin/sh PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin @reboot root touch "/boot/vmlinuz-"`uname -r`