This document deals with:
- Running an Urbit ship with the ordinary runtime from the command line.
- Basic setup, configuration and usage in Urbit's shell called the
You can turn your urbit off with
Ctrl-d from the Chat or Dojo prompts.
You can force-quit your urbit with
Ctrl-z from anywhere.
To restart your urbit simply pass the name of your pier:
$ ./urbit some-planet
$ ./urbit comet
To log an urbit's command line output to a file, use
$ script urbit.log ./urbit your-urbit
Moving your pier
Piers are designed to be portable, but it must be done while the urbit is not running. Urbit networking is stateful, so you can't run two copies of the same urbit in two places.
To move a pier, simply move the contents of the directory it lives in.
To keep these files as small as possible we usually use the
tar. With a pier
your-urbit/, something like this should work:
tar -Scvzf ~/your-urbit.tar.gz ~/your-urbit/ scp your-old-server:~/your-urbit.tar.gz your-new-server:~
Then to unzip it, on your other Unix server, run:
tar xfvz your-urbit.tar.gz
Delete the tar file, and, after installing Urbit on your new server, start your urbit back up with:
Urbit can run on any x86 computer (unofficial, unsupported ARM binaries are also available), ideally with at least 2GB of RAM.
Urbit maintains a persistent log of the history of your ship. Eventually this log will be automatically trimmed when necessary, but for now it only increases in size. An actively used planet will consume 5-50 GB of storage space per year of operation.
Your Urbit terminal is separated into two parts: the prompt (the bottom line) and the record (everything above that). The record is shared; all the output from all the apps in your command set appears in it. The prompt is multiplexed.
In the CLI, Urbit apps can process your input before you hit return. To see this in action try entering
) as the first character at the Dojo prompt. Since there is no Dojo command or Hoon expression that starts with ')', the Dojo rejects it.
Ctrl-x - Switches the prompt between running console apps
Ctrl-c - Crash current event. Processed at the Unix layer and prints a stack
Ctrl-d - From Chat or Dojo, stops your Urbit process.
Ctrl-z - Stops the Urbit process from anywhere.
↓ - History navigation
The following emacs-style key bindings are available:
Ctrl-a Cursor to beginning of the line (Home) Ctrl-b Cursor one character backward (left-arrow) Ctrl-e Cursor to the end of the line (End) Ctrl-f Cursor one character forward (right-arrow) Ctrl-g Beep; cancel reverse-search Ctrl-k Kill to end of line Ctrl-l Clear the screen Ctrl-n Next line in history (down-arrow) Ctrl-p Previous line in history (up-arrow) Ctrl-r Reverse-search Ctrl-t Transpose characters Ctrl-u Kill to beginning of line Ctrl-y Yank from kill buffer
By default, your
%base desk (which contains the Arvo kernel and core apps) receives updates (OTAs) from your sponsor. Other desks will receive updates from their respective publishers. To check the OTA source for each desk, run
+vats in the dojo. It will print out details for each desk - the
source field shows which ship the desk gets updates from and the
updates field shows
tracking if automatic updates are enabled.
If for some reason updates are not enabled or the current source is not online or up to date, you can enable updates or change source with the
|install (sein:title our now our) %landscape will enable updates to the
%landscape desk from your sponsor.
|install ~some-ship %landscape will enable updates to the landscape desk from whatever ship is specified in place of
~some-ship. For third party apps, make sure to correctly specify the publisher's ship. Each desk's updates are managed separately, so you'll need to run this for each desk separately. For the
%base desk specifically, you sync from
%kids rather than
%base on the remote ship, so must specify it like
|install (sein:title our now our) %kids, =local %base.
Additional OTA Troubleshooting
Please check the Support Wiki for additional OTA troubleshooting, such as: OTA 1.0.71 failed, Missing OTA, Stuck flow preventing planets from receiving OTAs, and No content shows in Links page after OTA.
On startup, urbit tries to bind to
localhost:80. If you're already running something on port
80, or your host OS will not allow urbit to bind port
80, urbit will try
8082, and so on. For planets only, we also provide subdomains of
arvo.network for free. Any planet
~your-urbit is also at
your-urbit.arvo.network, but only after you set up DNS.
Once running, you can sign into your ship’s web interface from
http://localhost (if bound to port
http://localhost:8080 (if bound to port
https://your-urbit.arvo.network if you've set up DNS.
Planets can spawn moons, which are conceptually meant for connected devices: phones, smart TVs, digital thermostats. The basic idea is that your planet runs permanently in a data center somewhere, while moons run on all your devices. Each planet can issue ~4 billion (
To generate a random moon from your planet, run:
~sampel-palnet:dojo> |moon moon: ~faswep-navred-sampel-palnet 0w5cT5t.wCO6i.~e1xg.Oz0qb.QNO6I.3Kt2T.h9M9F.U3vU~.X3Qu0.gtb19.IYTkY.80kWZ.SIEUE.DXa8i.TiDof.o3-1C.RHLKS.k81M0.ecz5o.ic0Bg.600g1
moon: part is the name of the moon, in this case
~faswep-navred-sampel-palnet. The next line starting with
0w5... is the private key necessary to boot it.
You can just copy the key (which in this case would be the
0w5[...]600g1 part) to the clipboard, or save it in a
.key file, for example
You can use the key and moon name in the same installation flow from the Command line installation guide, following the same scheme as for booting a planet. That scheme is:
$ ./urbit -w <moon-name> -G <key> -c <pier-name>
$ ./urbit -w <moon-name> -k <key-file> -c <pier-name>
excludes the leading
-c argument is not required, but it is recommended; otherwise, the resulting directory is a rather unwieldy moon name. Moons are automatically synced to their parent
%kids desk, and can control applications on their parent planet using
To breach a moon -- that is, to reset its presence on the network so that it's treated as a freshly spawned ship by others -- run from the parent ship:
To cycle the keys of a moon without breaching, run:
You can then run
|rekey on the moon with the key given by the above command as the argument.
Escaping A Sponsor
To use the network as a planet or star, you must be sponsored by an active star or galaxy, respectively. If your sponsor isn't suiting your needs, you can escape to a different one. This can be done with Bridge following the instructions here.
Life and rift number
You can check your ship's life and rift number by running
+keys our in
dojo. You can inspect another ship's life and rift number by running
+keys ~sampel-palnet. For information on what life and rift are, see Life and Rift.
We have a system that lets you request a domain name for your ship in the form of
ship is your ship's name minus the
~. This allows users to access their ships remotely using Landscape, our graphical web interface. Stars and planets follow the same DNS request process, and galaxies have their own requirements. Moons and comets are not supported.
For a planet or star's DNS request to be made and fulfilled, they must be hosting their ship someplace with a public IP address, and its HTTP server must be listening on port 80.
To initiate a DNS request, run the following thread in your ship's dojo, passing the IP address as an argument with .0.0.0.0 (
@if) syntax. For example:
-dns-address [%if .126.96.36.199]
%dns-address thread, running locally, will make an HTTP request to that IP address on port 80 to confirm that it is itself available at that IP and port. If that fails, you'll receive a
couldn't access ship on port 80 message in the terminal; this request will retry a few times. If the self-check is successful, the request is relayed to
~zod, and you'll receive a message saying,
request for DNS sent to ~zod. Once
~zod has acknowledged receipt of the request, the
%dns-address thread will print a terminal message saying
awaiting response from ~zod.
The request will make take a little time to be fulfilled, but eventually the
ship.arvo.network DNS record will be set to the given IP address. Once that's set up,
~zod will be notified and
~zod will, in turn, notify your ship. That ship will now try to verify that it can reach itself on
ship.arvo.network over port 80. If it can't, it'll send a message saying,
unable to access via ship.arvo.network. If it can, it will configure itself with that domain and say
confirmed access via ship.arvo.network.
Great! You're set up now. Try accessing your
ship.arvo.network in your browser to use Landscape; we recommend Chrome or Brave.
To enable SSL on your ship, you must poke the
%acme agent with the domain encoded in a path and it will request a certificate. The path format is
/tld/your_domain/your_subdomain, so if your domain is
sampel-palnet.arvo.network, you'd use it like so:
:acme &path /network/arvo/sampel-palnet
Galaxies are already required to have separate DNS entry at galaxy.urbit.org. There's no automated process for getting that binding, so if you're a galaxy-holder, get in touch with us at [email protected].
There is a command for galaxies that will try to re-use their already-necessary Ames DNS entry for HTTPS:
This will make HTTP-requests to self-check availability over
galaxy.$AMES-DOMAIN (currently galaxy.urbit.org), where
galaxy is the galaxy's name minus the
-dns-auto works the same as
-dns-address does with stars and planets: if it's available or unavailable, terminal messages, and so on.
The built-in logic for listening on port 80 is to try to bind to port 80; if it cannot, it tries 8080, then increments until it can bind a port. Port 80 is available to unprivileged process on recent versions of macOS. Otherwise, the process needs to either be run as root, or be given special permission (
sudo setcap 'cap_net_bind_service=+ep' /path/to/urbit/binary on Linux).