Ship Troubleshooting

Urbit is still in the development stage, so there's a chance that your ship won't start properly, or will stop working properly when you're running it. That's ok! This document is intended to help you in such an event.

Table of Contents

Best Practices

An ounce of prevention is worth a pound of cure, so let's first go over some best practices to keep your ship in working order.

Only boot with your keyfile once

Once your ship is booted with your keyfile, you should never use that same keyfile again. If you do boot with the same keyfile twice, any other ship on the network that your ship has communicated with will never be able to talk to it again.

If you accidentally booted with the same keyfile twice, the only remedy is performing a personal breach, which is explained in the next section.

Do not delete your pier

Urbit is stateful, meaning that it needs to hold onto all your data. If you delete your pier and start your ship again, you won't be able to talk to any ship you've talked to before. The only solution to this is performing a personal breach

Keep track of the directory that you put your ship in

When you first start your ship, you should make sure you put it a place where you can find it again and where it won't get accidentally deleted. Remember that you must perform |mount % in your ship's Dojo to make your ship visible as a directory in the Unix file system.

Avoid killing the Urbit process directly

The best way to end an urbit process is to use ctrl-d from the Dojo. Unix methods to kill the process, such as with ctrl-z or with the kill Bash command, or simply closing the window, should only be used if ctrl-d does not work.

Keep up-to-date builds

Check for latest Urbit version at https://github.com/urbit/urbit/releases. If you're behind, update using this guide

|hi your star to see if you're connected

Find out who your star is by running (sein:title our now our) in the Dojo. Then, run |hi ~star, where ~star is the star's name, and if things are working properly, you should get the message hi ~star successful. It could also be helpful to use |hi to check connectivity with ~zod or another planet that you're in a Talk channel with.

Turn your ship off and on again

Use ctrl-d to gracefully exit your ship, and then start it again. This can solve many issues.

Use the |knob command to customize your error messages

Error messages can be by overwhelming, so the |knob command is intended to remedy this. It's used to silence errors that aren't important.

The command takes two arguments, and comes in the form of |knob %error-tag %level.

%error-tag is the name of the error in question. It's usually printed at the top of the stack trace, such as in crud: %hole event failed -- %hole here is an example of an error tag.

%level determines how much you will see of errors with your chosen %error-tag. There are three levels:

So for example, to silence all Ames packet-related errors, try |knob %hole %hush.

Perform a personal breach.

A personal breach is when a ship tells all the other ships that have communicated with it to treat it as though the ship was just started for the first time again, since everyone has network has forgotten about it.

Personal breaches often fix connectivity issues, but should only be used as a last resort. To find out how to perform a personal breach, check out our Guide to Breaches.

Operation Issues

My urbit is very slow

Run |wash-gall. This clears caches in Gall, and may result in steep performance improvements.

My urbit is frozen

Sometimes this happens if you're processing a very large event, or if you're in an infinite loop, or for a variety of other reasons.

Before doing anything, try waiting for a minute: an event might finish processing. If it doesn't clear up, then use Unix kill-command, ctrl-z, to end your ship's process. Then restart your ship.

When I try to type into the Dojo, it prints dy-noprompt

This happens when your Dojo is waiting on a request, such as an HTTP request. You can fix it simply by typing backspace or (delete on Mac).

My ship doesn't recognize file changes that I make in my pier

Since version 0.8.0, changes no longer automatically sync between the Unix side (your pier) and your ship. To sync your file changes, you must run |commit %desk in your Dojo, where %desk is the desk you'd like to sync.

Connectivity Issues

I can |hi my sponsor star, but I cannot get syncs from it or discover peers

This means you have one-way connectivity with your star, with your star being unable to send your messages; many connectivity problems are derived from this. Run |nuke ~star, where ~star is your sponsor star in the Dojo. Then run |nuke ~star again. This can solve various communication problems. You can find your sponsor by running (sein:title our now our).

I can't communicate with anyone

You may have booted a ship with your keyfile twice in the same era. To fix this, you must perform a personal breach.

I keep getting an ames error stack-trace

You may see a message like this one: /~zod/home/~2019.7.22..18.55.46..83a3/sys/vane/ames:<[line column].[line column]>. This is a clay path to a Hoon file, pointing to the line and column where an expression crashed. This kind of error might be accompanied by a crud message.

This means that another ship is sending invalid packets to you. This could be because one of the ships has not updated the other ship's "life number," which is the number that starts at one and increments every time that ship performs a personal breach.

This can happen if they have the wrong keys of yours, or if you have the wrong keys of theirs. You can figure out who has the wrong keys by running this scry command in your dojo: .^(* %j /=life=/shipname), where shipname is the other ship's name. Save that information. Then, go to the Azimuth contract on Etherscan, scroll down to 32. points, and put in the hexadecimal representation of the other ship's @p. You can find the hexadecimal representation by running...

`@ux`~sampel-palnet

in the Dojo, where ~sampel-palnet is the other ship's name. Then, compare it to the scry information that you saved. If that information matches up, it means that the other ship is the problem. If it doesn't match up, your ship has wrong information about the other ship. If you have such wrong information, you can fix this by running:

:azimuth-tracker|listen ~ %app %azimuth-tracker

The last line above syncs from an Ethereum node for all ships on the network. If you only wanted to sync with certain ships, run:

:azimuth-tracker|listen ~[~sampel-palnet ~zod ~marzod] %app %azimuth-tracker

~sampel-palnet ~zod ~marzod are example ship-names; replace these with any number of desired ships-names.

The above commands work if you have the wrong keys of other ships. If other ships have wrong keys of yours, you need to somehow ask them to to run such a command.

I can talk to some ships, but I can't talk to my sponsor and some other ships

This is the result of deleting your pier and starting your ship again. To fix this, you must perform a personal breach.

I have two ships, and only one can connect with the other

One-way connectivity (OWC) means that ship ~A can't |hi ship ~B, but ship ~B can |hi ship ~A. If you have two ships that are suffering from this problem, it can be solved by a series of steps.

  1. Turn both ships off and on again. This sometimes solves this problem.

  2. On ship ~B -- the ship that can connect to the other -- |nuke ~A, where ~A is the ship that cannot |hi the other. Wait several minutes. Do you see any messages in your console? If not, try sending a |hi ~A from ~B. If “blocked” or similar messages start appearing, |nuke ~A again to unblock, and check if connectivity has been restored.

  3. If you can’t get any packets to be sent to ~B, you’re going to have to temporarily modify your ship’s Ames to trick it into sending packets. Make sure to do this on its home desk so as to not impact any syncs. Open ~A/home/sys/vanes/ames.hoon and find ++ harv. Just above (gth caw nif) check, insert the a line of this code: =? caw =(caw nif) +(caw). Save the file and run |commit %home.

  4. Try running |hi ~B. Wait a bit. Are you seeing %blocked or %bad-ack messages yet? If not, send another |hi ~B.

  5. Once the messages start rolling in, revert the Ames change that you made on ~A in step 3. Then |nuke ~A again from ~B. Wait a little bit, then check if this resolved the OWC.

  6. If the OWC is still not resolved, try step 6.

  7. Be patient. After following these steps, your |hi may still not go through instantly even though OWC has silently been fixed: there could be something like a very large sync queued up.

Booting Issues

My ship won't boot and gets a terminals database is inaccessible message

This happens when you try to run the Urbit binary through your $PATH. When running the command, you need to be explicit and specify the actual path to it in the filesystem. This is a side-effect of the way we build the release binaries, which will be fixed.

My ship booted for the first time, but it turned into a comet instead of my planet or star

You may have used the wrong arguments when booting your ship for the first time. Delete this comet and try again.

My development ship ("fakezod") gets a boot: malformed failure

This means that you gave your development ship an invalid @p. So, you will get this error if you write, for example, urbit -F zodzod instead of urbit -F zod.

Crashing Issues

I got a bail error and my ship crashed

Try just bringing it back up; it will often start working just fine again.

However, if you get a bail error again, you should perform a personal breach.

Making a GitHub issue out of your bail

You can get help with you problem by creating an issue at github.com/urbit/urbit. But to make a good issue, you need to include some information.

When your urbit crashes with a bail, you'll probably get a core dump, which is a file that contains the program state of your urbit when it crashed. On Mac, core dumps can be found in /cores. On Linux, cores can often be found in /var/crash, or the home directory.

Navigate to the folder containing your core dumps. Find the most recent core dump by looking at the dates after you run ls -l. Then lldb -c <corename>. Once that loads, you'll be at an (lldb) prompt; type bt at this prompt. This will create a stack trace that looks like this:

(lldb) bt
* thread #1, stop reason = signal SIGSTOP
  * frame #0: 0x000000010583d871 urbit`_box_free + 17
    frame #1: 0x0000000105845ee6 urbit`u3j_boot + 182
    frame #2: 0x000000010584d1f9 urbit`u3m_boot + 89
    frame #3: 0x000000010583d15d urbit`main + 2765
    frame #4: 0x00007fff75cb83d5 libdyld.dylib`start + 1
(lldb)

Copy this stack trace and include it in your GitHub issue.

My ship crashed with a bail: meme error.

  1. Make sure you are running version to 0.8.2 if you are not already on it.

  2. Restart your ship. If you don't crash again, everything may be fine. If you do crash again, then you should perform a personal breach.

My ship crashed with a bail: oops error

Restart your ship. These issues often just go away on their own. If this error repeats after restart two or more times, post the messages in an issue at github.com/urbit/urbit/issues.

This same error might also appear with a message like Assertion '0'.

My ship crashed with an pier: work error error

This means that the Urbit worker process has shut down for one reason or another. Just restart your ship; this is not a notable or reportable error.