Migrate a vBulletin 4 forum to Discourse

This guide covers migrating a vBulletin 4 forum to Discourse with the importer script at script/import_scripts/vbulletin.rb.

:warning: Expect to adjust the script. Every vBulletin install has its own plugins, custom fields and BBCode quirks. Most real imports need small changes to the script. Do a full test import on a development machine before touching production.

Which importer should I use?

Forum Script
vBulletin 4 script/import_scripts/vbulletin.rb (this guide)
vBulletin 3 script/import_scripts/vbulletin3.rb (same steps, different script name)
vBulletin 5 script/import_scripts/vbulletin5.rb
Very large forums (millions of posts) script/bulk_import/vbulletin.rb (much faster, but has fewer features and is harder to customize)

What gets imported

  • Groups
  • Users (banned users become suspended users)
  • Forums: each parent and child forum becomes a category and subcategory
  • Topics
  • Sticky thread → pinned topic
  • Closed thread → closed topic
  • Soft-deleted or moderated thread → unlisted topic
  • Posts
  • Private messages
  • Attachments
  • Avatars (only if they’re stored in the database, see below)
  • Permalinks for topics

Caveats

  • You can resume an interrupted import and re-run it against a newer database snapshot to pick up new content. Content that was already imported will not be updated with later edits.

  • Discourse supports two levels of categories. Forums nested more deeply are flattened.

  • BBCode lists ([ul], [ol], [li]) aren’t fully converted and may need manual cleanup.

Before you export

If vBulletin stores avatars in the file system, the importer won’t find them. Visit /admincp/avatar.php?do=storage on your vBulletin site and move avatars into the database before you export it.

Then export the database on the vBulletin server:

mysqldump -u USERNAME -p DATABASE_NAME | gzip > vb4_dump.sql.gz

Copy the dump and the attachments folder to the machine you’ll import on (for example, with scp or rsync). Your vBulletin admin settings show where the attachments are stored.

Option A: Import on a development machine (recommended)

Importing locally lets you iterate quickly and throw away failed attempts. When the result looks right, back up the site and restore it in production.

  1. Set up a Discourse development environment by following the Ubuntu/Debian or macOS guide.

  2. Install MariaDB. It’s a drop-in replacement for MySQL, and MySQL 5.7 packages are no longer available on current distributions.

    # Ubuntu/Debian
    sudo apt-get update
    sudo apt-get install mariadb-server libmariadb-dev
    sudo service mariadb start
    
     # macOS
     brew install mariadb
     brew services start mariadb
    
  3. Install the importer gems. mysql2 is included when IMPORT=1 is set. php-serialize needs to be added:

    cd ~/discourse
    echo "gem 'php-serialize', require: false" >> Gemfile
    IMPORT=1 bundle install
    
  4. Load the vBulletin database:

    sudo mysql -e "CREATE DATABASE vb4"
    gunzip < vb4_dump.sql.gz | sudo mysql vb4
    
  5. Start with an empty Discourse database:

    bin/rails db:drop db:create db:migrate
    
  6. Run the importer. You can stop and restart it at any time, because it skips content that’s already been imported.

    export DB_HOST="localhost"  
    export DB_NAME="vb4"
    export DB_USER="root"
    export DB_PW=""                                # MariaDB root password, if you set one
    export TABLE_PREFIX="vb_"                      # Your table prefix, or "" if there isn't one
    export ATTACHMENT_DIR="/path/to/attachments"   # The attachments folder you copied
    export TIMEZONE="America/Los_Angeles"          # The timezone vBulletin used
    
    IMPORT=1 bundle exec ruby script/import_scripts/vbulletin.rb
    
  7. Start Discourse and let background jobs finish. Run bin/dev, and in a second terminal run bundle exec sidekiq. Post rebaking and image processing can take a long time on a big forum. You can follow the progress at http://localhost:4200/sidekiq.

  8. Move to production. Make a backup and restore it on your production site by following Move a Discourse site to another VPS.

Option B: Import directly in the production container

  1. Install Discourse with the standard install guide.

  2. Adjust site settings (see Recommended site settings below).

  3. Copy the dump and attachments into /var/discourse/shared/standalone/ on the host. Inside the container they appear under /shared/, and files there survive a container rebuild.

  4. Enter the container and install MariaDB:

    cd /var/discourse
    ./launcher enter app
    apt-get update
    apt-get install -y mariadb-server libmariadb-dev
    service mariadb start
    
  5. Install the importer gems:

    cd /var/www/discourse
    echo "gem 'php-serialize', require: false" >> Gemfile
    su discourse -c 'bundle config set --local deployment false && bundle config set --local frozen false'
    su discourse -c 'IMPORT=1 bundle install'
    
  6. Load the vBulletin database:

    mysql -e "CREATE DATABASE vb4"
    gunzip < /shared/vb4_dump.sql.gz | mysql vb4
    
  7. Run the importer. Run it inside tmux or screen so an SSH disconnect doesn’t kill it.

    cd /var/www/discourse
    su discourse -c 'DB_NAME=vb4 DB_USER=root DB_PW="" TABLE_PREFIX=vb_ \
    ATTACHMENT_DIR=/shared/attachments TIMEZONE=America/Los_Angeles \
    IMPORT=1 bundle exec ruby script/import_scripts/vbulletin.rb'
    
  8. Clean up. Delete the files you copied into /shared, then rebuild the container. The rebuild removes MariaDB and the extra gems and restores the original Gemfile.

    exit
    cd /var/discourse
    ./launcher rebuild app
    

Recommended site settings

Set these before importing. You can also set them in get_site_settings_for_import in script/import_scripts/base.rb, which helps when you run test imports repeatedly.

  • login_required: enable this until the import is done and checked.

  • slug_generation_method: set to encoded if your forum isn’t in English, so topic URLs keep their non-ASCII titles.

  • download_remote_images_to_local: disable this if you don’t want Discourse to download images hotlinked in old posts.

  • disable_system_edit_notifications: enable this so users aren’t notified about every post the system edits while it processes images.

Known issues

  • Integer values are not supported (ArgumentError) in parse_timestamp. This affects older copies of the script running with newer versions of tzinfo. Update to the latest Discourse, or change `parse_timestamp` to:

    def parse_timestamp(timestamp)
    return nil if timestamp.blank?
       Time.zone.at(@tz.utc_to_local(Time.at(timestamp.to_i)))
    end
    
  • Quotes with usernames that have spaces, or nested quotes, aren’t converted. See this reply for improved regexes.

  • Can't connect to local server through socket. MariaDB isn’t running. Start it with service mariadb start.

21 Likes

The script will now import avatars, but they must be embedded into the database. To do so, I’m told that you visit /admincp/avatar.php?do=storage and do something there. It seemed to work for the last client I did an import for.

Hey, @erlend_sh, make this a wiki and I’ll add to it in a few days when I submit a PR for this importer.

6 Likes

I was following your guide to import my vBulletin forum to Discourse. I am using a docker container for Discourse. I ran the following command:

I got a long list of errors:

sudo: no tty present and no askpass program specified
sudo: no tty present and no askpass program specified
[...]
sudo: no tty present and no askpass program specified
sudo: no tty present and no askpass program specified

Bundler::SudoNotPermittedError: Bundler requires sudo access to install at the moment. Try installing again, granting Bundler sudo access when
prompted, or installing into a different path.
An error occurred while installing rake (11.2.2), and Bundler cannot continue.
Make sure that `gem install rake -v '11.2.2'` succeeds before bundling.

I am new to ruby on rails and dockers. I do understand the concept of gems and Bundler.
What do these errors mean and how can I resolve them?

1 Like

Ok so basically I had to install all the necessary gems.
The script worked quite well.

However, I am not able to login with one of the users I created on my vBulletin forum just before importing the database over to discourse.

1 Like

You’ll need to set passwords. The script doesn’t import them (last I looked).

And even if it did, you’d need to install a plugin to make them work.

1 Like

I am planning to write a script that will import the passwords for my case.
I know the mysql tables I need to consider for vBulletin. What is the table in PostgreSQL for Discourse that stores passwords and is used during login?

Prakarsh would you be able to share the forum url that you are trying to switch from? The vBulletin one. Will the Discourse one have the same URL?

Yes. Here it is: Brainstorm

Our plan is to keep the URL same for now. I cannot say when it will be live. Probably by end of next week if everything goes right :stuck_out_tongue:

1 Like

Before you do that, you should read this topic.

You don’t want to directly touch the Discourse tables. You want a script like this. When I last used it and tried to import the passwords, it ran really slowly. I’m not sure why, but after reading this I decided that it’s probably not a bad idea to make people change their passwords.

3 Likes

@pfaffman - You mentioned you used import_pass field in your import script.
Where exactly do you put the import_pass field? And how do you use it?
If you could share it would be great help.

I have tried using this script but it failed 90% into the import of the users:

su discourse -c 'RAILS_ENV=production ruby script/import_scripts/vbulletin.rb'
root:XXXX@localhost wants vb4
loading existing groups...
loading existing users...
loading existing categories...
loading existing posts...
loading existing topics...

importing groups...
        6 / 19 ( 31.6%)  Failed to create group id 7 Moderators: ["Name has already been taken"]
       19 / 19 (100.0%)  
importing users
     2746 / 15319 ( 17.9%)  W, [2017-10-12T10:07:05.104837 #3578]  WARN -- : Bad date/time value "0000:00:00 00:00:00": mon out of range
W, [2017-10-12T10:07:05.106562 #3578]  WARN -- : Bad date/time value "0000:00:00 00:00:00": mon out of range
W, [2017-10-12T10:07:05.107305 #3578]  WARN -- : Bad date/time value "0000:00:00 00:00:00": mon out of range
    13901 / 15319 ( 90.7%)  script/import_scripts/vbulletin.rb:137:in `strip': invalid byte sequence in UTF-8 (ArgumentError)
        from script/import_scripts/vbulletin.rb:137:in `block (2 levels) in import_users'
        from /var/www/discourse/script/import_scripts/base.rb:226:in `block in create_users'
        from /var/www/discourse/script/import_scripts/base.rb:225:in `each'
        from /var/www/discourse/script/import_scripts/base.rb:225:in `create_users'
        from script/import_scripts/vbulletin.rb:133:in `block in import_users'
        from /var/www/discourse/script/import_scripts/base.rb:784:in `block in batches'
        from /var/www/discourse/script/import_scripts/base.rb:783:in `loop'
        from /var/www/discourse/script/import_scripts/base.rb:783:in `batches'
        from script/import_scripts/vbulletin.rb:117:in `import_users'
        from script/import_scripts/vbulletin.rb:78:in `execute'
        from /var/www/discourse/script/import_scripts/base.rb:45:in `perform'
        from script/import_scripts/vbulletin.rb:902:in `<main>'

Anyone have any suggestions regarding how to debug and fix this?

You have an encoding problem in your database.

The database claims is using one encoding, but some of the data is encoded some other way. A job I just did had a similar problem. UTF-8 likely does not exist when you started your forum.

This is probably a problem for stack exchange or a database person. It’s not a problem with the importer.

It’s the kind of problem that will take 10 minutes following a recipe that you don’t understand, ten hours looking for that recipe, or 100 hours fixing it by hand.

3 Likes

Cheers @pfaffman I was approaching the same conclusion… I started writing up the following before your comment…

It looks like it might be caused by non-ascii characters in usernames, so dumping the usernames into a text file and searching for UTF-8:

echo "select username from user" | mysql -p vb4 | grep --color='auto' -P -n "[^[:ascii:]]"

These were the entities this threw up:

’ ´ – é ã Ñ í ó 

So I used this page to generate encoded versions and updated the usernames in the database, for example:

UPDATE user SET username="&rsquo; &acute; &ndash; &eacute; &atilde; &Ntilde; &iacute; &oacute;" WHERE username="’ ´ – é ã Ñ í ó"

But the script still fails:

script/import_scripts/vbulletin.rb:137:in `strip': invalid byte sequence in UTF-8 (ArgumentError)
        from script/import_scripts/vbulletin.rb:137:in `block (2 levels) in import_users'
        from /var/www/discourse/script/import_scripts/base.rb:226:in `block in create_users'
        from /var/www/discourse/script/import_scripts/base.rb:225:in `each'
        from /var/www/discourse/script/import_scripts/base.rb:225:in `create_users'
        from script/import_scripts/vbulletin.rb:133:in `block in import_users'
        from /var/www/discourse/script/import_scripts/base.rb:784:in `block in batches'
        from /var/www/discourse/script/import_scripts/base.rb:783:in `loop'
        from /var/www/discourse/script/import_scripts/base.rb:783:in `batches'
        from script/import_scripts/vbulletin.rb:117:in `import_users'
        from script/import_scripts/vbulletin.rb:78:in `execute'
        from /var/www/discourse/script/import_scripts/base.rb:45:in `perform'
        from script/import_scripts/vbulletin.rb:902:in `<main>'

So I then dumped the whole users table into a text file:

echo "select * from user" | mysql -p vb4 > /tmp/allusers.txt

And then tried all the suggesting in this stackexchange thread for finding non-ascii characters and none were found, so now to search the whole database for the problem…

When I find a solution I’ll post it here in case it helps others…

2 Likes

The problem almost certainly exists in other tables.

You need to figure out what it’s encoded in and fix it.

Or, it could be that for some years it’s in one encoding and other years it’s another.

Best of luck. I was relieved they my client had engineers they solved the problem so I didn’t have to. I may have some notes that I’ll try to find when I get out of bed.

1 Like

The database on the old server was UTF-8 but clearly it has some non-UTF-8 in it, the site has been in use for a long time and has 13k users… so taking care with dumping the old database:

mysqldump -uroot -p database -r dump.sql

And trying various methods to check the encoding (thanks to Stackexchange):

file dump.sql 
  dump.sql: ASCII text, with very long lines

apt install uchardet 
uchardet dump.sql
  UTF-8

iconv -f utf-8 -t utf-8 dump.sql > dump.utf-8.sql
  iconv: illegal input sequence at position 29618109

apt install moreutils
isutf8 dump.sql 
  dump.sql: line 2995, char 42, byte 29618109: Expecting bytes in the following ranges: 00..7F C2..F4.

And this is a image line:

INSERT INTO `customavatar` VALUES (6355,'ÿØÿà\0^PJFIF\0^A^A\0\0^A\0^A\0\0ÿþ\0;CREATOR: gd-jpeg v1.0 (using IJG JPEG v62), quality = 75

And before I delete the line using vim I need to increase the size of the partitions as it is a 1.4G file and I’m running out of space when editing it…

1 Like

The problem was a user with &#55357;&#56444; in the username field, once the HTML entities were removed the import script ran, but there were a lot of errors like this:

ERR Error running script (call to f_b06356ba4628144e123b652c99605b873107c9be): @user_script:14: @user_script: 14: -MISCONF Redis is configured to save RDB snapshots, but is currently not able to persist on disk. Commands that may modify the data set are disabled. Please check Redis logs for details about the error. 

Is this something that I should be concerned about and something that I should fix and then redo the import?

Are you out of disk space?

1 Like

Yes! I’m now spinning up a new virtual server with a huge amount of disk space, CPUs and RAM and I’m going to start from scratch…

By the way, I’m following the instructions above for the Docker version and found that I needed to add an additional step in the container

echo "discourse ALL = NOPASSWD: ALL" >> /etc/sudoers

Before running:

su discourse -c 'bundle install --no-deployment --without test --without development'

In the non-docker instructions above there is this step to delete data from the database before doing the import:

  • Clear existing data from your local Discourse instance
cd ~/discourse
bundle exec rake db:drop db:create db:migrate

Was this omitted from the Docker instructions on purpose?

1 Like

There are two, different, vbulletin import scripts, in the docker container:

cd /var/www/discourse
find ./ -name vbulletin.rb
./script/bulk_import/vbulletin.rb
./script/import_scripts/vbulletin.rb

The second one is twice the size of the first one and it is the one I have been using.