Vagrant environment setup tutorial¶
This section guides first-time contributors through installing the Zulip development environment on Windows, macOS, Ubuntu and Debian.
The recommended method for installing the Zulip development environment is to use Vagrant with VirtualBox on Windows and macOS, and Vagrant with Docker on Ubuntu. This method creates a virtual machine (for Windows and macOS) or a Linux container (for Ubuntu) inside which the Zulip server and all related services will run.
If you encounter errors installing the Zulip development environment, check Troubleshooting and Common Errors. If that doesn’t help, please visit #provision help in the Zulip development community server for real-time help or file an issue.
When reporting your issue, please include the following information:
host operating system
installation method (Vagrant or direct)
whether or not you are using a proxy
a copy of Zulip’s
vagrantprovisioning logs, available in
/var/log/provision.logon your virtual machine
Installing the Zulip development environment with Vagrant requires downloading several hundred megabytes of dependencies. You will need an active internet connection throughout the entire installation processes. (See Specifying a proxy if you need a proxy to access the internet.)
All: 2GB available RAM, Active broadband internet connection, GitHub account.
macOS: macOS (10.11 El Capitan or newer recommended)
Ubuntu LTS: 18.04 or 16.04 64-bit
or Debian: 9.0 “stretch” 64-bit
Windows: Windows 64-bit (Win 10 recommended), hardware virtualization enabled (VT-x or AMD-V), administrator access.
Other Linux distributions work great too, but we don’t maintain documentation for installing Vagrant and Docker on those systems, so you’ll need to find a separate guide and crib from the Debian/Ubuntu docs.
Step 0: Set up Git & GitHub¶
You can skip this step if you already have Git, GitHub, and SSH access to GitHub working on your machine.
Follow our Git Guide in order to install Git, set up a GitHub account, create an SSH key to access code on GitHub efficiently, etc. Be sure to create an ssh key and add it to your GitHub account using these instructions.
Step 1: Install Prerequisites¶
Now you are ready for Step 2: Get Zulip Code..
1. Install Vagrant, Docker, and Git¶
christie@ubuntu-desktop:~ $ sudo apt install vagrant docker.io git
2. Add yourself to the
christie@ubuntu-desktop:~ $ sudo adduser $USER docker Adding user `christie' to group `docker' ... Adding user christie to group docker Done.
You will need to reboot for this change to take effect. If it worked,
you will see
docker in your list of groups:
christie@ubuntu-desktop:~ $ groups | grep docker christie adm cdrom sudo dip plugdev lpadmin sambashare docker
3. Make sure the Docker daemon is running:¶
If you had previously installed and removed an older version of Docker, an Ubuntu bug may prevent Docker from being automatically enabled and started after installation. You can check using the following:
$ systemctl status docker ● docker.service - Docker Application Container Engine Loaded: loaded (/lib/systemd/system/docker.service; enabled; vendor preset: enabled) Active: active (running) since Mon 2019-07-15 23:20:46 IST; 18min ago
If the service is not running, you’ll see
Active: inactive (dead) on
the second line, and will need to enable and start the Docker service
using the following:
sudo systemctl unmask docker sudo systemctl enable docker sudo systemctl start docker
Now you are ready for Step 2: Get Zulip Code.
Install Git for Windows, which installs Git BASH.
Install VirtualBox (latest).
Install Vagrant (latest).
(Note: While Git BASH is recommended, you may also use Cygwin. If you do, make sure to install default required packages along with git, curl, openssh, and rsync binaries.)
Also, you must have hardware virtualization enabled (VT-x or AMD-V) in your computer’s BIOS.
Running Git BASH as an administrator¶
It is important that you always run Git BASH with administrator privileges when working on Zulip code, as not doing so will cause errors in the development environment (such as symlink creation). You might wish to configure your Git BASH shortcut to always run with these privileges enabled (see this guide for how to set this up).
Step 2: Get Zulip Code¶
In your browser, visit https://github.com/zulip/zulip and click the
forkbutton. You will need to be logged in to GitHub to do this.
Open Terminal (macOS/Ubuntu) or Git BASH (Windows; must run as an Administrator).
git clone --config pull.rebase email@example.com:YOURUSERNAME/zulip.git cd zulip git remote add -f upstream https://github.com/zulip/zulip.git
This will create a ‘zulip’ directory and download the Zulip code into it.
Don’t forget to replace YOURUSERNAME with your git username. You will see something like:
christie@win10 ~ $ git clone --config pull.rebase firstname.lastname@example.org:YOURUSERNAME/zulip.git Cloning into 'zulip'... remote: Counting objects: 73571, done. remote: Compressing objects: 100% (2/2), done. remote: Total 73571 (delta 1), reused 0 (delta 0), pack-reused 73569 Receiving objects: 100% (73571/73571), 105.30 MiB | 6.46 MiB/s, done. Resolving deltas: 100% (51448/51448), done. Checking connectivity... done. Checking out files: 100% (1912/1912), done.`
Now you are ready for Step 3: Start the development environment.
Step 3: Start the development environment¶
Change into the zulip directory and tell vagrant to start the Zulip
development environment with
# On Windows or macOS: cd zulip vagrant plugin install vagrant-vbguest vagrant up --provider=virtualbox # On Linux: cd zulip vagrant up --provider=docker
The first time you run this command it will take some time because vagrant does the following:
downloads the base Ubuntu 18.04 virtual machine image (for macOS and Windows) or container (for Ubuntu)
configures this virtual machine/container for use with Zulip,
creates a shared directory mapping your clone of the Zulip code inside the virtual machine/container at
tools/provisionscript inside the virtual machine/container, which downloads all required dependencies, sets up the python environment for the Zulip development server, and initializes a default test database. We call this process “provisioning”, and it is documented in some detail in our dependencies documentation.
You will need an active internet connection during the entire
process. (See Specifying a proxy if you need a
proxy to access the internet.)
vagrant up can fail while
provisioning if your Internet connection is unreliable. To retry, you
vagrant provision (
vagrant up will just boot the guest
without provisioning after the first time). Other common issues are
documented in the
Troubleshooting and Common Errors
section. If that doesn’t help, please visit
in the Zulip development community server for
On Windows, you will see
The system cannot find the path specified. message
several times. This is expected behavior and is not an error.
vagrant up has completed, connect to the development
christie@win10 ~/zulip $ vagrant ssh
You should see output that starts like this:
Welcome to Ubuntu 18.04.2 LTS (GNU/Linux 4.15.0-54-generic x86_64)
Congrats, you’re now inside the Zulip development environment!
You can confirm this by looking at the command prompt, which starts
(zulip-py3-venv)vagrant@. If it just starts with
provisioning failed and you should look at the
Next, start the Zulip server:
(zulip-py3-venv) vagrant@ubuntu-bionic:/srv/zulip $ ./tools/run-dev.py
You will see several lines of output starting with something like:
2016-05-04 22:20:33,895 INFO: process_fts_updates starting Recompiling templates 2016-05-04 18:20:34,804 INFO: Not in recovery; listening for FTS updates done Validating Django models.py... System check identified no issues (0 silenced). Django version 1.8 Tornado server is running at http://localhost:9993/ Quit the server with CTRL-C. 2016-05-04 18:20:40,716 INFO Tornado loaded 0 event queues in 0.001s 2016-05-04 18:20:40,722 INFO Tornado 95.5% busy over the past 0.0 seconds Performing system checks...
And ending with something similar to:
http://localhost:9994/webpack-dev-server/ webpack result is served from http://localhost:9991/webpack/ content is served from /srv/zulip webpack: bundle is now VALID. 2016-05-06 21:43:29,553 INFO Tornado 31.6% busy over the past 10.6 seconds 2016-05-06 21:43:35,007 INFO Tornado 23.9% busy over the past 16.0 seconds
Now the Zulip server should be running and accessible. Verify this by navigating to http://localhost:9991/ in the browser on your main machine.
You should see something like this:
The Zulip server will continue to run and send output to the terminal window. When you navigate to Zulip in your browser, check your terminal and you should see something like:
2016-05-04 18:21:57,547 INFO 127.0.0.1 GET 302 582ms (+start: 417ms) / (unauth via ?) [04/May/2016 18:21:57]"GET / HTTP/1.0" 302 0 2016-05-04 18:21:57,568 INFO 127.0.0.1 GET 301 4ms /login (unauth via ?) [04/May/2016 18:21:57]"GET /login HTTP/1.0" 301 0 2016-05-04 18:21:57,819 INFO 127.0.0.1 GET 200 209ms (db: 7ms/2q) /login/ (unauth via ?)
Now you’re ready for Step 4: Developing.
Step 4: Developing¶
Where to edit files¶
You’ll work by editing files on your host machine, in the directory where you cloned Zulip. Use your favorite editor (Sublime, Atom, Vim, Emacs, Notepad++, etc.).
When you save changes they will be synced automatically to the Zulip development environment on the virtual machine/container.
Each component of the Zulip development server will automatically restart itself or reload data appropriately when you make changes. So, to see your changes, all you usually have to do is reload your browser. More details on how this works are available below.
Zulip’s whitespace rules are all enforced by linters, so be sure to
tools/lint often to make sure you’re following our coding style
tools/setup-git-repo to run it on just the changed files
automatically whenever you commit).
Understanding run-dev.py debugging output¶
It’s good to have the terminal running
run-dev.py up as you work since error
messages including tracebacks along with every backend request will be printed
See Logging for further details on the run-dev.py console output.
Committing and pushing changes with git¶
When you’re ready to commit or push changes via git, you will do this by running git commands in Terminal (macOS/Ubuntu) or Git BASH (Windows) in the directory where you cloned Zulip on your main machine.
If you’re new to working with Git/GitHub, check out our Git & GitHub Guide.
Maintaining the development environment¶
If after rebasing onto a new version of the Zulip server, you receive
new errors while starting the Zulip server or running tests, this is
probably not because Zulip’s master branch is broken. Instead, this
is likely because we’ve recently merged changes to the development
environment provisioning process that you need to apply to your
development environment. To update your environment, you’ll need to
re-provision your vagrant machine using
vagrant provision (this just
tools/provision from your Zulip checkout inside the Vagrant
guest); this should complete in about a minute.
After provisioning, you’ll want to (re)start the Zulip development server.
Rebuilding the development environment¶
If you ever want to recreate your development environment again from
scratch (e.g. to test as change you’ve made to the provisioning
process, or because you think something is broken), you can do so
vagrant destroy and then
vagrant up. This will usually be
much faster than the original
vagrant up since the base image is
already cached on your machine (it takes about 5 minutes to run with a
fast Internet connection).
Any additional programs (e.g. Zsh, emacs, etc.) or configuration that
you may have installed in the development environment will be lost
when you recreate it. To address this, you can create a script called
tools/custom_provision in your Zulip Git checkout; and place any
extra setup commands there. Vagrant will run
every time you run
vagrant provision (or create a Vagrant guest via
Shutting down the development environment for use later¶
To shut down but preserve the development environment so you can use
it again later use
vagrant halt or
You can do this from the same Terminal/Git BASH window that is running
run-dev.py by pressing ^C to halt the server and then typing
exit. Or you
can halt vagrant from another Terminal/Git BASH window.
From the window where run-dev.py is running:
2016-05-04 18:33:13,330 INFO 127.0.0.1 GET 200 92ms /register/ (unauth via ?) ^C KeyboardInterrupt (zulip-py3-venv) vagrant@ubuntu-bionic:/srv/zulip$ exit logout Connection to 127.0.0.1 closed. christie@win10 ~/zulip
Now you can suspend the development environment:
christie@win10 ~/zulip $ vagrant suspend ==> default: Saving VM state and suspending execution...
vagrant suspend doesn’t work, try
christie@win10 ~/zulip $ vagrant halt ==> default: Attempting graceful shutdown of VM...
Resuming the development environment¶
When you’re ready to work on Zulip again, run
vagrant up (no need to
--provider option required above). You will also need to
connect to the virtual machine with
vagrant ssh and re-start the
christie@win10 ~/zulip $ vagrant up $ vagrant ssh (zulip-py3-venv) vagrant@ubuntu-bionic:/srv/zulip $ ./tools/run-dev.py
Next, read the following to learn more about developing for Zulip:
Troubleshooting and Common Errors¶
Below you’ll find a list of common errors and their solutions. Most
issues are resolved by just provisioning again (by running
/srv/zulip) inside the Vagrant guest or
vagrant provision from outside).
If these solutions aren’t working for you or you encounter an issue not documented below, there are a few ways to get further help:
When reporting your issue, please include the following information:
host operating system
installation method (Vagrant or direct)
whether or not you are using a proxy
a copy of Zulip’s
vagrantprovisioning logs, available in
/var/log/provision.logon your virtual machine. If you choose to post just the error output, please include the beginning of the error output, not just the last few lines.
The output of
tools/diagnose run inside the Vagrant guest is also
Vagrant guest doesn’t show (zulip-py3-venv) at start of prompt¶
This is caused by provisioning failing to complete successfully. You
can see the errors in
var/log/provision.log; it should end with
something like this:
ESC[94mZulip development environment setup succeeded!ESC[0m
ESC stuff are the terminal color codes that make it show as a nice
blue in the terminal, which unfortunately looks ugly in the logs.
If you encounter an incomplete
/var/log/provision.log file, you need to
update your environment. Re-provision your vagrant machine; if the problem
persists, please come chat with us (see instructions above) for help.
After you provision successfully, you’ll need to exit your
shell and run
vagrant ssh again to get the virtualenv setup properly.
ssl read error¶
If you receive the following error while running
SSL read: error:00000000:lib(0):func(0):reason(0), errno 104
It means that either your network connection is unstable and/or very
slow. To resolve it, run
vagrant up until it works (possibly on a
better network connection).
Unmet dependencies error¶
vagrant up or
provision, if you see the following error:
==> default: E:unmet dependencies. Try 'apt-get -f install' with no packages (or specify a solution).
It means that your local apt repository has been corrupted, which can usually be resolved by executing the command:
apt-get -f install
ssh connection closed by remote host¶
vagrant ssh, if you see the following error:
ssh_exchange_identification: Connection closed by remote host
It usually means the Vagrant guest is not running, which is usually
solved by rebooting the Vagrant guest via
vagrant halt; vagrant up. See
Vagrant was unable to communicate with the guest machine
for more details.
Hyper-V error messages¶
If you get an error message on Windows about lack of Windows Home
support for Hyper-V when running
vagrant up, the problem is that
Windows is incorrectly attempting to use Hyper-V rather than
Virtualbox as the virtualization provider. You can fix this by
explicitly passing the virtualbox provider to
christie@win10 ~/zulip $ vagrant up --provide=virtualbox
Connection timeout on
If you see the following error after running
default: SSH address: 127.0.0.1:2222 default: SSH username: vagrant default: SSH auth method: private key default: Error: Connection timeout. Retrying... default: Error: Connection timeout. Retrying... default: Error: Connection timeout. Retrying...
A likely cause is that hardware virtualization is not enabled for your computer. This must be done via your computer’s BIOS settings. Look for a setting called VT-x (Intel) or (AMD-V).
If this is already enabled in your BIOS, double-check that you are running a 64-bit operating system.
For further information about troubleshooting vagrant timeout errors see this post.
Vagrant was unable to communicate with the guest machine¶
If you see the following error when you run
Timed out while waiting for the machine to boot. This means that Vagrant was unable to communicate with the guest machine within the configured ("config.vm.boot_timeout" value) time period. If you look above, you should be able to see the error(s) that Vagrant had when attempting to connect to the machine. These errors are usually good hints as to what may be wrong. If you're using a custom box, make sure that networking is properly working and you're able to connect to the machine. It is a common problem that networking isn't setup properly in these boxes. Verify that authentication configurations are also setup properly, as well. If the box appears to be booting properly, you may want to increase the timeout ("config.vm.boot_timeout") value.
This has a range of possible causes, that usually amount to a bug in
Virtualbox or Vagrant. If you see this error, you usually can fix it
by rebooting the guest via
vagrant halt; vagrant up.
Vagrant up fails with subprocess.CalledProcessError¶
vagrant up command basically does the following:
Downloads an Ubuntu image and starts it using a Vagrant provider.
vagrant sshto connect to that Ubuntu guest, and then runs
tools/provision, which has a lot of subcommands that are executed via Python’s
subprocessmodule. These errors mean that one of those subcommands failed.
To debug such errors, you can log in to the Vagrant guest machine by
vagrant ssh, which should present you with a standard shell
prompt. You can debug interactively by using e.g.
cd zulip && ./tools/provision, and then running the individual subcommands
that failed. Once you’ve resolved the problem, you can rerun
tools/provision to proceed; the provisioning system is designed
to recover well from failures.
The zulip provisioning system is generally highly reliable; the most common cause of issues here is a poor network connection (or one where you need a proxy to access the Internet and haven’t configured the development environment to use it.
Once you’ve provisioned successfully, you’ll get output like this:
Zulip development environment setup succeeded! (zulip-py3-venv) vagrant@vagrant-base-trusty-amd64:~/zulip$
(zulip-py3-venv) part is missing, this is because your
installation failed the first time before the Zulip virtualenv was
created. You can fix this by just closing the shell and running
vagrant ssh again, or using
Finally, if you encounter any issues that weren’t caused by your Internet connection, please report them! We try hard to keep Zulip development environment provisioning free of bugs.
pip install fails during
vagrant up on Ubuntu¶
Likely causes are:
Insufficient RAM. Check whether you’ve allotted at least two gigabytes of RAM, which is the minimum Zulip requires. If not, go to your VM settings and increase the RAM, then restart the VM.
yarn install warnings¶
$ yarn install yarn install v0.24.5 [1/4] Resolving packages... [2/4] Fetching packages... warning email@example.com: The platform "linux" is incompatible with this module. info "firstname.lastname@example.org" is an optional dependency and failed compatibility check. Excluding it from installation. [3/4] Linking dependencies... [4/4] Building fresh packages... $ browserify node_modules/sockjs-client/lib/entry.js --standalone SockJS > node_modules/sockjs-client/sockjs.js Done in 23.50s.
OSError: [Errno 26] Text file busy¶
default: Traceback (most recent call last): … default: File "/srv/zulip-py3-venv/lib/python3.6/shutil.py", line 426, in _rmtree_safe_fd default: os.rmdir(name, dir_fd=topfd) default: OSError: [Errno 26] Text file busy: 'baremetrics'
This error is caused by a bug in recent versions of the VirtualBox Guest Additions for Linux on Windows hosts. It has not been fixed upstream as of this writing, but you may be able to work around it by removing the plugin that upgrades Guest Additions:
vagrant destroy vagrant plugin uninstall vagrant-vbguest vagrant up --provider=virtualbox
Specifying an Ubuntu mirror¶
Bringing up a development environment for the first time involves
downloading many packages from the Ubuntu archive. The Ubuntu cloud
images use the global mirror
default, but you may find that you can speed up the download by using
a local mirror closer to your location. To do this, create
~/.zulip-vagrant-config and add a line like this, replacing the URL
Specifying a proxy¶
If you need to use a proxy server to access the Internet, you will
need to specify the proxy settings before running
First, install the Vagrant plugin
vagrant plugin install vagrant-proxyconf
~/.zulip-vagrant-config and add the following lines to
it (with the appropriate values in it for your proxy):
HTTP_PROXY http://proxy_host:port HTTPS_PROXY http://proxy_host:port NO_PROXY localhost,127.0.0.1,.example.com,.zulipdev.com
For proxies that require authentication, the config will be a bit more complex, e.g.:
HTTP_PROXY http://userName:userPassword@192.168.1.1:8080 HTTPS_PROXY http://userName:userPassword@192.168.1.1:8080 NO_PROXY localhost,127.0.0.1,.example.com,.zulipdev.com
You’ll want to double-check your work for mistakes (a common one
https:// when your proxy expects
http://). Invalid proxy
configuration can cause confusing/weird exceptions; if you’re using a
proxy and get an error, the first thing you should investigate is
whether you entered your proxy configuration correctly.
vagrant up in your terminal to install the development
server. If you ran
vagrant up before and failed, you’ll need to run
vagrant destroy first to clean up the failed installation.
If you no longer want to use proxy with Vagrant, you can remove the
HTTPS_PROXY lines in
then do a
Using a different port for Vagrant¶
You can also change the port on the host machine that Vagrant uses by
adding to your
~/.zulip-vagrant-config file. E.g. if you set:
vagrant reload to apply the new configuration), then you would visit
http://localhost:9971/ to connect to your development server.
If you’d like to be able to connect to your development environment from other machines than the VM host, you can manually set the host IP address in the ‘~/.zulip-vagrant-config’ file as well. For example, if you set:
(and restart the Vagrant guest with
vagrant reload), your host IP would be
0.0.0.0, a special value for the IP address that means any IP address can
connect to your development server.
Customizing CPU and RAM allocation¶
When running Vagrant using a VM-based provider such as VirtualBox or VMWare Fusion, CPU and RAM resources must be explicitly allocated to the guest system (with Docker and other container-based Vagrant providers, explicit allocation is unnecessary and the settings described here are ignored).
Our default Vagrant settings allocate 2 cpus with 2GiB of memory for the guest, which is sufficient to run everything in the development environment. If your host system has more CPUs, or you have enough RAM that you’d like to allocate more than 2GiB to the guest, you can improve performance of the Zulip development environment by allocating more resources.
To do so, create a
~/.zulip-vagrant-config file containing the
GUEST_CPUS <number of cpus> GUEST_MEMORY_MB <system memory (in MB)>
GUEST_CPUS 4 GUEST_MEMORY_MB 8192
would result in an allocation of 4 cpus and 8 GiB of memory to the guest VM.
After changing the configuration, run
vagrant reload to reboot the
guest VM with your new configuration.
If at any time you wish to revert back to the default settings, simply
GUEST_MEMORY_MB lines from