Backups, export and import

Zulip has high quality export and import tools that can be used to move data from one Zulip server to another, do backups, compliance work, or migrate from your own servers to the hosted Zulip Cloud service (or back):

  • The Backup tool is designed for exact restoration of a Zulip server’s state, for disaster recovery, testing with production data, or hardware migration. This tool has a few limitations:

    • Backups must be restored on a server running the same Zulip version (most precisely, one where manage.py showmigrations has the same output).

    • Backups must be restored on a server running the same PostgreSQL version. To install Zulip with the same version of PostgreSQL that the backup was taken on, pass the desired version with the --postgresql-version argument when installing.

    • Backups aren’t useful for migrating organizations between self-hosting and Zulip Cloud (which may require renumbering all the users/messages/etc.).

    We highly recommend this tool in situations where it is applicable, because it is highly optimized and highly stable, since the hard work is done by the built-in backup feature of PostgreSQL. We also document backup details for users managing backups manually.

  • The logical Data export tool is designed for migrating data between Zulip Cloud and other Zulip servers, as well as various auditing purposes. The logical export tool produces a .tar.gz archive with most of the Zulip database data encoded in JSON files–a format shared by our data import tools for third-party services like Slack.

    Like the backup tool, logical data exports must be imported on a Zulip server running the same version. However, logical data exports can be imported on Zulip servers running a different PostgreSQL version or hosting a different set of Zulip organizations. We recommend this tool in cases where the backup tool isn’t applicable, including situations where an easily machine-parsable export format is desired.

  • Compliance exports allow a server administrator to export messages matching a search query.

  • Zulip also has an HTML archive tool, which is primarily intended for public archives, but can also be useful to inexpensively preserve public stream conversations when decommissioning a Zulip organization.

  • It’s possible to set up PostgreSQL streaming replication and the S3 file upload backend as part of a high availability environment.

Backups

The Zulip server has a built-in backup tool:

# As the zulip user
/home/zulip/deployments/current/manage.py backup
# Or as root
su zulip -c '/home/zulip/deployments/current/manage.py backup'

The backup tool provides the following options:

  • --output=/tmp/backup.tar.gz: Filename to write the backup tarball to (default: write to a file in /tmp). On success, the console output will show the path to the output tarball.

  • --skip-db: Skip backup of the database. Useful if you’re using a remote PostgreSQL host with its own backup system and just need to back up non-database state.

  • --skip-uploads: If LOCAL_UPLOADS_DIR is set, user-uploaded files in that directory will be ignored.

This will generate a .tar.gz archive containing all the data stored on your Zulip server that would be needed to restore your Zulip server’s state on another machine perfectly.

Restoring backups

  1. Install the same base OS as the backup was taken on. If you want to upgrade the OS, you should do this after restoring the backup.

  2. Install a new Zulip server through Step 3, with the same version of PostgreSQL that the backup was taken on, by passing the desired version with the --postgresql-version argument. If you want to upgrade the version of PostgreSQL, you should do this after restoring the backup.

  3. As root, import the backup:

/home/zulip/deployments/current/scripts/setup/restore-backup /path/to/backup

When that finishes, your Zulip server should be fully operational again.

Changing the hostname

It’s common, when testing backup restoration, to restore backups with a different user-facing hostname than the original server to avoid disrupting service (e.g. zuliptest.example.com rather than zulip.example.com).

If you do so, just like any other time you change the hostname, you’ll need to update EXTERNAL_HOST and then restart the Zulip server (after backup restoration completes).

Until you do, your Zulip server will think its user-facing hostname is still zulip.example.com and will return HTTP 400 BAD REQUEST errors when trying to access it via zuliptest.example.com.

Changing database settings

If you wish to restore onto a very differently configured host (e.g. with REMOTE_POSTGRES_HOST set to a different value), you can edit /etc/zulip/settings.py to configure the host to suit the new host’s needs, then restore with --keep-settings:

/home/zulip/deployments/current/scripts/setup/restore-backup --keep-settings /path/to/backup

You can also pass --keep-zulipconf if you wish to preserve the local /etc/zulip/zulip.conf.

Inspecting a backup tarball

If you’re not sure what versions were in use when a given backup was created, you can get that information via the files in the backup tarball: postgres-version, os-version, and zulip-version. The following command may be useful for viewing these files without extracting the entire archive.

tar -Oaxf /path/to/archive/zulip-backup-rest.tar.gz zulip-backup/zulip-version

What is included

Backups contain everything you need to fully restore your Zulip server, including the database, settings, secrets from /etc/zulip, and user-uploaded files stored on the Zulip server.

The following data is not included in these backup archives, and you may want to back up separately:

  • The server access/error logs from /var/log/zulip. The Zulip server only appends to logs, and they can be very large compared to the rest of the data for a Zulip server.

  • Files uploaded with the Zulip S3 file upload backend. We don’t include these for two reasons. First, the uploaded file data in S3 can easily be many times larger than the rest of the backup, and downloading it all to a server doing a backup could easily exceed its disk capacity. Additionally, S3 is a reliable persistent storage system with its own high-quality tools for doing backups.

  • SSL certificates. These are not included because they are particularly security-sensitive and are either trivially replaced (if generated via Certbot) or provided by the system administrator.

For completeness, Zulip’s backups do not include certain highly transient state that Zulip doesn’t store in a database. For example, typing status data, API rate-limiting counters, and RabbitMQ queues that are essentially always empty in a healthy server (like outgoing emails to send). You can check whether these queues are empty using rabbitmqctl list_queues.

Backup details

This section is primarily for users managing backups themselves (E.g. if they’re using a remote PostgreSQL database with an existing backup strategy), and also serves as documentation for what is included in the backups generated by Zulip’s standard tools. The data includes:

  • The PostgreSQL database. You can back this up with any standard database export or backup tool; see below for Zulip’s built-in support for continuous point-in-time backups.

  • Any user-uploaded files. If you’re using S3 as storage for file uploads, this is backed up in S3. But if you have instead set LOCAL_UPLOADS_DIR, any files uploaded by users (including avatars) will be stored in that directory and you’ll want to back it up.

  • Your Zulip configuration including secrets from /etc/zulip/. E.g. if you lose the value of secret_key, all users will need to log in again when you set up a replacement server since you won’t be able to verify their cookies. If you lose avatar_salt, any user-uploaded avatars will need to be re-uploaded (since avatar filenames are computed using a hash of avatar_salt and user’s email), etc.

Restore from manual backups

To restore from a manual backup, the process is basically the reverse of the above:

  • Install new server as normal by downloading a Zulip release tarball and then using scripts/setup/install. You should pass --no-init-db because we don’t need to create a new database.

  • Unpack to /etc/zulip the settings.py and zulip-secrets.conf files from your backups.

  • Restore your database from the backup.

  • Reconfigure rabbitmq to use the password from secrets.conf by running, as root, scripts/setup/configure-rabbitmq.

  • If you’re using local file uploads, restore those files to the path specified by settings.LOCAL_UPLOADS_DIR and (if appropriate) any logs.

  • Start the server using scripts/restart-server.

This restoration process can also be used to migrate a Zulip installation from one server to another.

We recommend running a disaster recovery after setting up your backups to confirm that your backups are working. You may also want to monitor that they are up to date using the Nagios plugin at: puppet/zulip/files/nagios_plugins/zulip_postgresql_backups/check_postgresql_backup.

Data export

Zulip’s powerful data export tool is designed to handle migration of a Zulip organization between different Zulip installations; as a result, these exports contain all non-transient data for a Zulip organization, with the exception of passwords and API keys.

We recommend using the backup tool if your primary goal is backups.

Upgrade if exporting for import into Zulip Cloud

If you are exporting data from a self-hosted version of Zulip for purposes of importing into Zulip Cloud, you should first upgrade your server to the zulip-cloud-current branch:

/home/zulip/deployments/current/scripts/upgrade-zulip-from-git zulip-cloud-current

It is not sufficient to be on the latest stable release, as zulip.com runs pre-release versions of Zulip that are often several months of development ahead of the latest release.

Preventing changes during the export

For best results, you’ll want to shut down access to the organization before exporting; so that nobody can send new messages (etc.) while you’re exporting data. There are two ways to do this:

  1. ./scripts/stop-server, which stops the whole server. This is preferred if you’re not hosting multiple organizations, because it has no side effects other than disabling the Zulip server for the duration.

  2. Pass --deactivate to ./manage export, which first deactivates the target organization, logging out all active login sessions and preventing all accounts from logging in or accessing the API. This is preferred for environments like Zulip Cloud where you might want to export a single organization without disrupting any other users, and the intent is to move hosting of the organization (and forcing users to re-log in would be required as part of the hosting migration anyway).

We include both options in the instructions below, commented out so that neither runs (using the # at the start of the lines). If you’d like to use one of these options, remove the # at the start of the lines for the appropriate option.

Export your Zulip data

Log in to a shell on your Zulip server as the zulip user. Run the following commands:

cd /home/zulip/deployments/current
# ./scripts/stop-server
# export DEACTIVATE_FLAG="--deactivate"   # Deactivates the organization
./manage.py export -r '' $DEACTIVATE_FLAG # Exports the data

(The -r option lets you specify the organization to export; '' is the default organization hosted at the Zulip server’s root domain.)

This will generate a tarred archive with a name like /tmp/zulip-export-zcmpxfm6.tar.gz. The archive contains several JSON files (containing the Zulip organization’s data) as well as an archive of all the organization’s uploaded files.

Import into a new Zulip server

  1. Install a new Zulip server, skipping Step 3 (you’ll create your Zulip organization via the data import tool instead).

    • Ensure that the Zulip server you’re importing into is running the same version of Zulip as the server you’re exporting from.

    • For exports created from Zulip Cloud (zulip.com), you need to upgrade to zulip-cloud-current, which represents the current version that Zulip Cloud is running; this is generally main delayed by a week or two. To upgrade to that:

      /home/zulip/deployments/current/scripts/upgrade-zulip-from-git zulip-cloud-current

      It is not sufficient to be on the latest stable release, as zulip.com runs pre-release versions of Zulip that are often several months of development ahead of the latest release.

    • Note that if your server has limited free RAM, you’ll want to shut down the Zulip server with ./scripts/stop-server while you run the import, since our minimal system requirements do not budget extra RAM for running the data import tool.

  2. If your new Zulip server is meant to fully replace a previous Zulip server, you may want to copy some settings from /etc/zulip to your new server to reuse the server-level configuration and secret keys from your old server. There are a few important details to understand about doing so:

    • Copying /etc/zulip/settings.py and /etc/zulip/zulip.conf is safe and recommended. Care is required when copying secrets from /etc/zulip/zulip-secrets.conf (details below).

    • If you copy zulip_org_id and zulip_org_key (the credentials for the mobile push notifications service), you should be very careful to make sure the no users had their IDs renumbered during the import process (this can be checked using manage.py shell with some care). The push notifications service has a mapping of which user_id values are associated with which devices for a given Zulip server (represented by the zulip_org_id registration). This means that if any user_id values were renumbered during the import and you don’t register a new zulip_org_id, push notifications meant for the user who now has ID 15 may be sent to devices registered by the user who had user ID 15 before the data export (yikes!). The solution is simply to not copy these settings and re-register your server for mobile push notifications if any users had their IDs renumbered during the logical export/import process.

    • If you copy the rabbitmq_password secret from zulip-secrets.conf, you’ll need to run scripts/setup/configure-rabbitmq as root to update your local RabbitMQ installation to use the password in your Zulip secrets file.

    • You will likely want to copy camo_key (required to avoid breaking certain links) and any settings you added related to authentication and email delivery so that those work on your new server.

    • Copying avatar_salt is not recommended, due to similar issues to the mobile push notifications service. Zulip will automatically rewrite avatars at URLs appropriate for the new user IDs, and using the same avatar salt (and same server URL) post import could result in issues with browsers caching the avatar image improperly for users whose ID was renumbered.

  3. Log in to a shell on your Zulip server as the zulip user. Run the following commands, replacing the filename with the path to your data export tarball:

cd ~
tar -xf /path/to/export/file/zulip-export-zcmpxfm6.tar.gz
cd /home/zulip/deployments/current
./manage.py import '' ~/zulip-export-zcmpxfm6
./scripts/start-server

This could take several minutes to run depending on how much data you’re importing.

Import options

The commands above create an imported organization on the root domain (EXTERNAL_HOST) of the Zulip installation. You can also import into a custom subdomain, e.g. if you already have an existing organization on the root domain. Replace the last two lines above with the following, after replacing <subdomain> with the desired subdomain.

./manage.py import <subdomain> ~/zulip-export-zcmpxfm6
./scripts/start-server

Logging in

Once the import completes, all your users will have accounts in your new Zulip organization, but those accounts won’t have passwords yet (since for security reasons, passwords are not exported). Your users will need to either authenticate using something like Google auth or start by resetting their passwords.

You can use the ./manage.py send_password_reset_email command to send password reset emails to your users. We recommend starting with sending one to yourself for testing:

./manage.py send_password_reset_email -u username@example.com

and then once you’re ready, you can email them to everyone using e.g.

./manage.py send_password_reset_email -r '' --all-users

(replace '' with your subdomain if you’re using one).

Deleting and re-importing

If you did a test import of a Zulip organization, you may want to delete the test import data from your Zulip server before doing a final import. You can permanently delete all data from a Zulip organization by running (replacing '' with the subdomain if you are hosting the organization on a subdomain):

./manage.py delete_realm -r ''

Assuming you’re using the local file uploads backend, you can additionally delete all file uploads, avatars, and custom emoji on a Zulip server (across all organizations) with the following command:

rm -rf /home/zulip/uploads/*/*

If you’re hosting multiple organizations and would like to remove uploads from a single organization, you’ll need to access realm.id in the management shell before deleting the organization from the database (this will be 2 for the first organization created on a Zulip server, shown in the example below), e.g.:

rm -rf /home/zulip/uploads/*/2/

Once that’s done, you can simply re-run the import process.

Compliance exports

In some circumstances, corporate or legal compliance may require performing selective data exports. This can be done with the export_search command-line tool, which lets you specify the following parameters when exporting messages:

  • Search keywords in the message text.

  • Message sender or recipient.

  • Time range for when messages were sent.

For example, to search for messages containing the word “wonderland” between November 1st and 6th, from alice@example.com:

$ /home/zulip/deployments/current/manage.py export_search --output compliance-export.json
    -r zulip \
    --after '2022-11-01 00:00:00' --before '2022-11-06 14:00:00' \
    --sender alice@example.com \
    wonderland

The results are written to a JSON or CSV file. The contents of previous versions of edited messages are not searched, nor are deleted messages. Attachments associated with the resulting messages can optionally also be exported.

See /home/zulip/deployments/current/manage.py export_search --help for more details on supported options.

Database-only backup tools

The Zulip-specific backup tool documented above is perfect for an all-in-one backup solution, and can be used for nightly backups. For administrators wanting continuous point-in-time backups, Zulip has built-in support for taking daily backup snapshots along with streaming write-ahead log (WAL) backups using wal-g and storing them in Amazon S3. By default, these backups are stored for 30 days.

Note these database backups, by themselves, do not constitute a full backup of the Zulip system! See above for other pieces which are necessary to back up a Zulip system.

To enable continuous point-in-time backups:

  1. Edit /etc/zulip/zulip-secrets.conf on the PostgreSQL server to add:

s3_region = # region to write to S3; defaults to EC2 host's region
s3_backups_key = # aws public key; optional, if access not through role
s3_backups_secret_key =  # aws secret key; optional, if access not through role
s3_backups_bucket = # name of S3 backup bucket

  1. Run /home/zulip/deployments/current/scripts/zulip-puppet-apply.

Daily full-database backups will be taken at 0200 UTC, and every WAL archive file will be written to S3 as it is saved by PostgreSQL; these are written every 16KiB of the WAL. This means that if there are periods of slow activity, it may be minutes before the backup is saved into S3 – see archive_timeout for how to set an upper bound on this. On an active Zulip server, this also means the Zulip server will be regularly sending PutObject requests to S3, possibly thousands of times per day.

If you need always-current backup availability, Zulip also has built-in database replication support.

You can (and should) monitor that backups are running regularly via the Nagios plugin installed into /usr/lib/nagios/plugins/zulip_postgresql_backups/check_postgresql_backup.