Administrator Guide

This guide will cover how to perform common DoltLab administrator configuration and tasks for versions <= v2.0.8 which do not use the installer binary.

1. Backup DoltLab data

2. Authenticate a Dolt client to use DoltLab account

3. Monitor DoltLab with cAdvisor and Prometheus

4. Connect to an SMTP server with implicit TLS

5. Troubleshoot SMTP server connection problems

6. Prevent unauthorized user account creation

7. Use an external database server with DoltLab

8. Expose DoltLab on a closed host with ngrok

9. DoltLab Jobs

10. Disable usage metrics

11. Migrate old format DoltLab databases

12. Use a domain name with DoltLab

13. Run DoltLab on Hosted Dolt

14. Serve DoltLab over HTTPS with a TLS reverse proxy

15. Serve DoltLab over HTTPS natively

16. Improve DoltLab performance

17. Serve DoltLab behind an AWS Network Load Balancer

Backup and restore volumes

DoltLab persists all data to local disk using Docker volumes. To backup or restore DoltLab's data, we recommend the following steps which follow Docker's official volume backup and restore documentation, with the exception of DoltLab's PostgreSQL server.

DoltLab <= v0.8.4 uses PostgreSQL as its database and DoltLab v1.0.0+ uses Dolt. To backup the PostgreSQL server we recommend dumping the database with pg_dump and restoring the database from the dump using psql. To backup the Dolt server we recommend using Docker's volume backup and restore process, or Dolt's built-in backup and restore features.

Backing up and restoring remote data, user uploaded data, and Dolt server data with Docker

To backup DoltLab's remote data, the database data for all database on a given DoltLab instance, leave DoltLab's services up and run:

# backup remote data stored in DoltLab RemoteAPI's volume and save to a tar file
docker run --rm --volumes-from doltlab_doltlabremoteapi_1 -v $(pwd):/backup ubuntu tar cvf /backup/remote-data.tar /doltlab-remote-storage

This will create a tar file called remote-data.tar in your working directory.

To backup user uploaded files, run:

# backup remote data stored in DoltLab RemoteAPI's volume and save to a tar file
docker run --rm --volumes-from doltlab_doltlabfileserviceapi_1 -v $(pwd):/backup ubuntu tar cvf /backup/user-uploaded-data.tar /doltlab-user-uploads

This will create a tar file called user-uploaded-data.tar in your working directory.

To backup Dolt server data, run:

# backup Dolt's root volume
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu tar cvf /backup/doltlabdb-root.tar /.dolt

# backup Dolt's config volume
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu tar cvf /backup/doltlabdb-configs.tar /etc/dolt

# backup Dolt's data volume
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu tar cvf /backup/doltlabdb-data.tar /var/lib/dolt

# backup Dolt's local backup volume
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu tar cvf /backup/doltlabdb-backups.tar /backups

Before restoring DoltLab's volumes from a backup, first, stop the running DoltLab services, prune the Docker containers, and remove the old volume(s):

cd doltlab

# stop the DoltLab services
docker-compose stop

# prune containers
docker container prune

# remove the remote data volume
docker volume rm doltlab_doltlab-remote-storage

# remove the user uploaded data
docker volume rm doltlab_doltlab-user-uploads

# remove the Dolt server root volume
docker volume rm doltlab_doltlabdb-dolt-root

# remove the Dolt server config volume
docker volume rm doltlab_doltlabdb-dolt-configs

# remove the Dolt server data volume
docker volume rm doltlab_doltlabdb-dolt-data

# remove the Dolt server local backups volume
docker volume rm doltlab_doltlabdb-dolt-backups

Next, start DoltLab's services using the start-doltlab.sh script. After the script completes, stop DoltLab once more with docker-compose stop. Doing this will recreate the required containers so that their volumes can be updated with the commands below.

Once the services are stopped, cd into the directory containing the remote-data.tar backup file and run:

# restore remote data from tar
docker run --rm --volumes-from doltlab_doltlabremoteapi_1 -v $(pwd):/backup ubuntu bash -c "cd /doltlab-remote-storage && tar xvf /backup/remote-data.tar --strip 1"

To restore user uploaded data, cd into the directory containing user-uploaded-data.tar and run:

# restore remote data from tar
docker run --rm --volumes-from doltlab_doltlabfileserviceapi_1 -v $(pwd):/backup ubuntu bash -c "cd /doltlab-user-uploads && tar xvf /backup/user-uploaded-data.tar --strip 1"

To restore Dolt server root data, cd into the directory containing doltlabdb-root.tar and run:

# restore Dolt server root data from tar
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu bash -c "cd /.dolt && tar xvf /backup/doltlabdb-root.tar --strip 1"

To restore Dolt server config data, cd into the directory containing doltlabdb-configs.tar and run:

# restore Dolt server config data from tar
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu bash -c "cd /etc/dolt && tar xvf /backup/doltlabdb-configs.tar --strip 2"

To restore Dolt server data, cd into the directory containing doltlabdb-data.tar and run:

# restore Dolt server data from tar
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu bash -c "cd /var/lib/dolt && tar xvf /backup/doltlabdb-data.tar --strip 3"

To restore Dolt server local backup data, cd into the directory containing doltlabdb-backups.tar and run:

# restore Dolt server local backups data from tar
docker run --rm --volumes-from doltlab_doltlabdb_1 -v $(pwd):/backup ubuntu bash -c "cd /backups && tar xvf /backup/doltlabdb-backups.tar --strip 1"

You can now restart DoltLab, and should see all data restored from the tar files.

Backing up and restoring PostgreSQL data

For DoltLab versions <= v0.8.4, to backup data from DoltLab's postgres server, we recommend executing a data dump with pg_dump. To do so, keep DoltLab's services up and run:

For DoltLab v0.7.0 and later, use --network doltlab below.

# dump postgres to postgres-dump.sql
docker run --rm --network doltlab_doltlab -e PGPASSWORD=<POSTGRES_PASSWORD> -v $(pwd):/doltlab-db-dumps postgres:13-bullseye bash -c "pg_dump --host=doltlab_doltlabdb_1 --port=5432 --username=dolthubadmin dolthubapi > /doltlab-db-dumps/postgres-dump.sql"

The value of PGPASSWORD should be the POSTGRES_PASSWORD set when DoltLab was first deployed.

To restore a postgres server from postgres-dump.sql, first stop the running DoltLab services, remove the stopped containers, and remove the old postgres server volume:

cd doltlab

# stop DoltLab
docker-compose stop

# remove containers
docker container prune

# remove old postgres data volume
docker volume rm doltlab_doltlabdb-data

Next, edit the postgres-dump.sql file by adding the line SET session_replication_role = replica; near the top of the file:

SET statement_timeout = 0;
SET lock_timeout = 0;
SET idle_in_transaction_session_timeout = 0;
SET client_encoding = 'UTF8';
SET standard_conforming_strings = on;
SELECT pg_catalog.set_config('search_path', '', false);
SET check_function_bodies = false;
SET xmloption = content;
SET client_min_messages = warning;
SET row_security = off;

/* We add this to disable triggers and rules during import */
SET session_replication_role = replica;
...

Start DoltLab's services again using the start-doltlab.sh script. After the script completes, cd into the directory containing the postgres-dump.sql file and run:

For DoltLab v0.7.0 and later, use --network doltlab below.

# import the postgres dump into the running server
docker run --rm --network doltlab_doltlab -e PGPASSWORD=<POSTGRES_PASSWORD> -v $(pwd):/doltlab-db-dumps postgres:13-bullseye bash -c "psql --host=doltlab_doltlabdb_1 --port=5432 --username=dolthubadmin dolthubapi < /doltlab-db-dumps/postgres-dump.sql"

Backing up and restoring the Dolt server with dolt backup

DoltLab v1.0.0+ uses Dolt as its database. To back up the Dolt server of DoltLab using Dolt's built-in backup and restore features, keep DoltLab's services up and open a connection to the database. The quickest way to do this is with the ./shell-db.sh script included with DoltLab:

DOLT_PASSWORD=<DOLT_PASSWORD> ./shell-db.sh
...
mysql>

Next, add a local backup using the DOLT_BACKUP() stored procedure. By default, DoltLab uses a Docker volume backed by the host's disk that allows you to create backups of the Dolt server. These backups will be located at /backups from within the Dolt server container. To create persistent backups, simply use /backups as the path prefix to the backup names:

mysql> call dolt_backup('add', 'local-backup', 'file:///backups/dolthubapi/2023/06/01');
+---------+
| success |
+---------+
|       1 |
+---------+
1 row in set (0.00 sec)

The above snippet will create a new backup stored at /backups/dolthubapi/2023/06/01 within the Dolt server container, and persisted to the host using the Docker volume doltlab_doltlabdb-dolt-backups.

You can sync the backup with the sync command:

mysql> call dolt_backup('sync', 'local-backup');
+---------+
| success |
+---------+
|       1 |
+---------+
1 row in set (0.00 sec)

The local backup is now synced, and you can now disconnect the shell.

At the time of this writing, Dolt only supports restoring backups using the CLI. To restore the Dolt server from a local backup, stop DoltLab's services using docker-compose stop. Then, use the ./dolt_db_cli.sh included with DoltLab to open a container shell with access to the Dolt server volumes.

Delete the existing ./dolthubapi directory located at /var/lib/dolt from within this container:

./dolt_db_cli.sh
root:/var/lib/dolt# ls
dolthubapi
root:/var/lib/dolt# rm -rf dolthubapi/

Doing this removes the existing Dolt server database. Now, use dolt backup restore to restore the database from the backup located at /backups/dolthubapi/2023/06/01:

root:/var/lib/dolt# dolt backup restore file:///backups/dolthubapi/2023/06/01 dolthubapi

root:/var/lib/dolt# ls
dolthubapi

The database has now been successfully restored, and you can now restart DoltLab.

Authenticate a Dolt client to use a DoltLab account

As of Dolt v0.39.0, the dolt login command can be used to authenticate against DoltLab instances.

To authenticate a client against DoltLab with this command, use the --auth-endpoint, --login-url, and --insecure arguments to point your Dolt client at the DoltLab instance you want to authenticate against.

--auth-endpoint should point at the DoltLab RemoteAPI Server running on port 50051. --login-url should point at the DoltLab instance's credentials page. --insecure a boolean flag, should be used since DoltLab does not currently support TLS gRPC connections.

dolt login --insecure --auth-endpoint doltlab.dolthub.com:50051 --login-url http://doltlab.dolthub.com/settings/credentials

Running the command will open your browser window to the --login-url with credentials populated in the "Public Key" field. Simply add a "Description" and click "Create", then return to your terminal to see your Dolt client successfully authenticated.

Credentials created successfully.
pub key: 9pg8rrkaqouuno4rihdlkb6pvaf16t134dscb68dlg0u0261rmtg
/.dolt/creds/cepi47m5sojotm8s2e8rkqm9fdjocidsc42canggvb8e0.jwk
Opening a browser to:
	http://doltlab.dolthub.com/settings/credentials#9pg8rrkaqouuno4rihdlkb6pvaf16t134dscb68dlg0u0261rmtg
Please associate your key with your account.
Checking remote server looking for key association.
requesting update
requesting update



Key successfully associated with user: <user> email <email>

For Dolt clients < v0.39.0, or to authenticate without using the dolt login command, first run the dolt creds new command, which will output a new public key:

dolt creds new
Credentials created successfully.
pub key: fef0kj7ia389i5atv8mcb31ksg9h3i6cji7aunm4jea9tccdl2cg

Copy the generated public key and run the dolt creds use command:

dolt creds use fef0kj7ia389i5atv8mcb31ksg9h3i6cji7aunm4jea9tccdl2cg

Next, login to your DoltLab account, click your profile image, then click "Settings" and then "Credentials".

Paste the public key into the "Public Key" field, write a description in the "Description" field, then click "Create".

Your Dolt client is now authenticated for this DoltLab account.

Monitor DoltLab with cAdvisor and Prometheus

As of DoltLab v0.3.0, Prometheus gRPC service metrics for DoltLab's Remote API Server, doltlabremoteapi, and DoltLab's Main API server, doltlabapi, are published on port 7770.

The metrics endpoints for these services are available at endpoints corresponding to their container name. For DoltLab's Remote API, thats :7770/doltlabremoteapi, and for DoltLab's Main API that's :7770/doltlabapi.

You can view the doltlabremoteapi service metrics for our DoltLab demo instance here, http://doltlab.dolthub.com:7770/doltlabremoteapi and you can view the doltlabapi service metrics here http://doltlab.dolthub.com:7770/doltlabapi.

To make these endpoints available to Prometheus, open port 7770 on your DoltLab host.

We recommend monitoring DoltLab with cAdvisor, which will expose container resource and performance metrics to Prometheus. Before running cAdvisor, open port 8080 on your DoltLab host as well. cAdvisor will display DoltLab's running containers via a web UI on :8080/docker and will publish Prometheus metrics for DoltLab's container at :8080/metrics by default.

Run cAdvisor as a Docker container in daemon mode with:

docker run -d -v /:/rootfs:ro -v /var/run:/var/run:rw -v /sys:/sys:ro -v /var/lib/docker/:/var/lib/docker:ro -v /dev/disk/:/dev/disk:ro -p 8080:8080 --name=cadvisor --privileged gcr.io/cadvisor/cadvisor:v0.39.3

To run a Prometheus server on your DoltLab host machine, first open port 9090 on the DoltLab host. Then, write the following prometheus.yml file on the host:

global:
  scrape_interval: 5s
  evaluation_interval: 10s
scrape_configs:
  - job_name: cadivsor
    static_configs:
      - targets: ["host.docker.internal:8080"]
  - job_name: prometheus
    static_configs:
      - targets: ["localhost:9090"]
  - job_name: doltlabremoteapi
    metrics_path: "/doltlabremoteapi"
    static_configs:
      - targets: ["host.docker.internal:7770"]
  - job_name: doltlabapi
    metrics_path: "/doltlabapi"
    static_configs:
      - targets: ["host.docker.internal:7770"]

Then, start the Prometheus server as a Docker container running in daemon mode:

docker run -d --add-host host.docker.internal:host-gateway --name=prometheus -p 9090:9090 -v "$(pwd)"/prometheus.yml:/etc/prometheus/prometheus.yml:ro prom/prometheus:latest --config.file=/etc/prometheus/prometheus.yml

--add-host host.docker.internal:host-gateway is only required if running the Prometheus server on the DoltLab host. If running it elsewhere, this argument may be omitted, and the host.docker.internal hostname in prometheus.yml can be changed to the hostname of your DoltLab host.

Connect to an SMTP server with implicit TLS

For DoltLab >= v0.4.2 and < v2.1.0, connections to existing SMTP servers using implicit TLS (on port 465, for example) are supported. To connect using implicit TLS, edit the docker-compose.yaml included in the DoltLab zip. Under the doltlabapi section, in the command block, add the following argument:

...
doltlabapi:
  ...
  command:
    ...
    -emailImplicitTLS
    ...

After adding the argument, restart DoltLab for it to take effect. Additionally, TLS verification can be skipped by adding the additional argument -emailInsecureTLS.

Troubleshoot SMTP server connection problems

DoltLab requires a connection to an existing SMTP server in order for users to create accounts, verify email addresses, reset forgotten passwords, and collaborate on databases.

Starting with DoltLab v0.4.1, the default user admin is created when DoltLab starts up, which allows admins to sign-in to their DoltLab instance even if they are experiencing SMTP server connection issues.

To help troubleshoot and resolve SMTP server connection issues, we've published the following go tool to help diagnose the SMTP connection issues on the host running DoltLab.

Starting with DoltLab v0.4.2, this tool is now included as an executable binary in DoltLab's zip, called smtp_connection_helper.

For usage run ./smtp_connection_helper --help which will output:

'smtp_connection_helper' is a simple tool used to ensure you can successfully connect to an smtp server.
If the connection is successful, this tool will send a test email to a single recipient from a single sender.
By default 'smtp_connection_helper' will attempt to connect to the SMTP server with STARTTLS. To use implicit TLS, use --implicit-tls

Usage:

./smtp_connection_helper \
--host <smtp hostname> \
--port <smtp port> \
--from <email address> \
--to <email address> \
--message {This is a test email message sent with smtp_connection_helper!} \
--subject {Testing SMTP Server Connection} \
--client-hostname {localhost} \
--auth <plain|login|external|anonymous|oauthbearer|disable> \
[--username smtp username] \
[--password smtp password] \
[--token smtp oauth token] \
[--identity smtp identity] \
[--trace anonymous trace] \
[--implicit-tls] \
[--insecure]

To send a test email using plain authentication, run:

./smtp_connection_helper \
--host existing.smtp.server.hostname \
--port 587 \ #STARTTLS port
--auth plain \
--username XXXXXXXX \
--password YYYYYYY \
--from [e-mail] \
--to [e-mail]
Sending email with auth method: plain
Successfully sent email!

To send a test email using plain authentication with implicit TLS, run:

./smtp_connection_helper \
--host existing.smtp.server.hostname \
--port 465 \ #TLS Wrapper Port
--implicit-tls \
--auth plain \
--username XXXXXXXX \
--password YYYYYYY \
--from [e-mail] \
--to [e-mail]
Sending email with auth method: plain
Successfully sent email!

To quickly get up and running with an existing SMTP server, we recommend using Gmail's. Once you've created a Gmail account, navigate to your account page and click the Security tab. Under the section "How you sign in to Google", click 2-Step Verification. If you have not yet setup 2-Step Verification, follow the prompts on this page to enable it.

After 2-Step Verification is set up, at the bottom of the page click "App passwords". Select app Mail and select a device, then click "Generate" to generate a password. This generated password can be supplied along with your Gmail email address to send emails with smtp_connection_helper and DoltLab.

./smtp_connection_helper \
--host smtp.gmail.com \
--port 587 \
--auth plain \
--username [e-mail] \
--password <generated App password> \
--from [e-mail] \
--to [e-mail]
Sending email with auth method: plain
Successfully sent email!

Prevent unauthorized user account creation

DoltLab for non-enterprise use currently supports explicit email whitelisting to prevent account creation by unauthorized users.

In DoltLab < v2.1.0, to enable DoltLab's email whitelisting feature, edit the docker-compose.yaml file included in DoltLab's zip.

Under the doltlabapi section, in the command block, remove the argument -dolthubWhitelistAllowAll. Restart your DoltLab instance for this to take effect.

Once DoltLab is restarted, your DoltLab instance will now check a PostgreSQL table called email_whitelist_elements before permitting account creation. Only user's with email addresses present in this table will be able to create accounts on your DoltLab instance.

To whitelist an email for account creation in your instance, you will need to insert their email address into the email_whitelist_elements table.

As of DoltLab v0.4.2, a script to easily connect to your DoltLab instance's running PostgreSQL server is included in the zip, called shell-db.sh.

Use this script by supplying the POSTGRES_PASSWORD you used to start your DoltLab instance, as PGPASSWORD here. Run:

PGPASSWORD=<your postgres password> ./shell-db.sh

You will see a dolthubapi=# PostgresSQL prompt connected to your DoltLab instance.

You can now execute the following INSERT to allow a specific user with [e-mail] to create an account on your DolLab instance:

INSERT INTO email_whitelist_elements (email_address, updated_at, created_at) VALUES ('[e-mail]', CURRENT_TIMESTAMP, CURRENT_TIMESTAMP);

For DoltLab >= v1.0.0 the same script can be used to update the email_whitelist_elements table, but expects DOLT_PASSWORD in place of PGPASSWORD. It will also open a mysql> prompt.

Use an external database server with DoltLab

For DoltLab v0.8.4 and earlier, you can connect a DoltLab instance to an external PostgreSQL server version 13 or later. To connect, in DoltLab's docker-compose.yaml, supply the host and port for the external server to doltlabapi's -pghost and -pgport arguments.

  doltlabapi:
    ...
    command:
      ...
      -pghost <host>
      -pgport <port>
      ...

You can also remove the doltlabdb section and all references to it and doltlabdb-data in the docker-compose.yaml file.

Before (re)starting DoltLab with this change, you will also need to execute the following statements in your external PostgreSQL server:

CREATE ROLE dolthubapi WITH LOGIN PASSWORD '$DOLTHUBAPI_PASSWORD';
CREATE DATABASE dolthubapi;
GRANT ALL PRIVILEGES ON DATABASE dolthubapi TO dolthubapi;
CREATE EXTENSION citext SCHEMA public;
GRANT ALL PRIVILEGES ON ALL TABLES IN SCHEMA public TO dolthubapi;

For DoltLab v1.0.0 or later, you can connect a DoltLab instance to an external Dolt server version v1.0.0 or later. To connect, in DoltLab's docker-compose.yaml, supply the host and port for the external server to doltlabapi's -doltHost and -doltPort arguments.

  doltlabapi:
    ...
    command:
      ...
      -doltHost <host>
      -doltPort <port>
      ...

Like with the external PostgreSQL changes described above, you can remove the doltlabdb section and all references to it and doltlabdb-dolt-data, doltlabdb-dolt-root, doltlabdb-dolt-configs, and doltlabdb-dolt-backups in the docker-compose.yaml file.

Finally, before (re)starting DoltLab with this change, you will also need to execute the following statements in your external Dolt server:

CREATE USER 'dolthubadmin' IDENTIFIED BY '<password>';
CREATE USER 'dolthubapi' IDENTIFIED BY '<password>';
GRANT ALL ON *.* TO 'dolthubadmin';
GRANT ALL ON dolthubapi.* TO 'dolthubapi';

Expose a DoltLab instance with ngrok

As of DoltLab v0.5.5, DoltLab instances can be exposed with ngrok. "How to expose DoltLab with ngrok" contains the instructions for this process, however, we do not recommend doing this for production DoltLab instances. This process requires one of DoltLab's services to be run without authentication, which may expose sensitive data. Do this at your own risk.

DoltLab Jobs

Jobs were introduced on DoltHub and are now available on DoltLab ^v0.7.0. DoltLab Jobs are stand-alone, long-running Docker containers that perform specific tasks for DoltLab users behind the scenes.

As a result of the Jobs infrastructure, DoltLab now requires more memory and disk. The amount of each of these depends on how users will use your instance.

We recommend running DoltLab on a host with at least 64 GB of memory, and 20 TBs of disk. These recommended amounts will decrease as we continue to improve Dolt's resource efficiency and utilization.

Disable usage metrics

By default, DoltLab collects first-party metrics for deployed instances. We use DoltLab's metrics to determine how many resources to allocate toward its development and improvement.

As of v0.7.0, DoltLab does not collect third-party metrics, and additionally, DoltLab's first-party metrics can be disabled. To disable metrics, edit the start-doltlab.sh script and remove run_with_metrics from the _main function.

Migrate old format DoltLab databases

Unlike DoltHub, DoltLab does not support automatic database migration for old format Dolt databases. Instead, old format database hosted on DoltLab need to be migrated manually. To migrate a DoltLab database:

1. Create a new database on DoltLab.

2. Clone the database you want to migrate.

3. Run dolt migrate in the cloned database.

4. Add the remote of the new DoltLab database to the cloned database with dolt remote add <remote name> http://<host ip>:50051/<owner>/<new db name>.

5. Push the migrated clone to the new database with dolt push <remote name> <branch name>.

Use a domain name with DoltLab

It's common practice to provision a domain name to use for a DoltLab instance. To do so, secure a domain name and map it to the stable, public IP address of the DoltLab host. Then, supply the domain name as the value to the HOST_IP environment variable when starting DoltLab. Let's look at an example using services offered by AWS.

Let's say we've have set up and run an EC2 instance with the latest version of DoltLab and have successfully configured its Security Group to allow ingress traffic on 80, 100, 4321, and 50051. By default, this host will have a public IP address assigned to it, but this IP is unstable and will change whenever the host is restarted.

First, we should attach a stable IP to this host. To do this in AWS, we can provision an Elastic IP Address (EIP).

Next, we should associate the EIP with our DoltLab host by following these steps. Once this is done, the DoltLab host should be reachable by the EIP.

Finally, we can provision a domain name for the DoltLab host through AWS Route 53. After registering the new domain name, we need to create an A record that's attached to the EIP of the DoltLab host. To do so, follow the steps for creating records outlined here.

Your DoltLab host should now be accessible via your new domain name. You can now stop your DoltLab server and replace the value of the environment variable HOST_IP with the domain name, then restart DoltLab.

In the event you are configuring your domain name with an Elastic Load Balancer, ensure that it specifies Target Groups for each of the ports required to operate DoltLab, 80, 100, 4321, and 50051.

Run DoltLab on Hosted Dolt

Starting with DoltLab v1.0.0, DoltLab can be configured to use a Hosted Dolt instance as its application database. This allows DoltLab administrators to use the feature-rich SQL workbench Hosted Dolt provides to interact with their DoltLab database.

To configure a DoltLab to use a Hosted Dolt, follow the steps below as we create a sample DoltLab Hosted Dolt instance called my-doltlab-db-1.

Create a Hosted Dolt deployment

To begin, you'll need to create a Hosted Dolt deployment that your DoltLab instance will connect to. We've created a video tutorial for how to create your first Hosted Dolt deployment, but briefly, you'll need to create an account on hosted.doltdb.com and then click the "Create Deployment" button.

You will then see a form where you can specify details about the host you need for your DoltLab instance:

In the image above you can see that we defined our Hosted Dolt deployment name as my-doltlab-db-1, selected an AWS EC2 host with 2 CPU and 8 GB of RAM in region us-west-2. We've also requested 200 GB of disk. For DoltLab, these settings should be more than sufficient.

We have also requested a replica instance by checking the "Enable Replication" box, and specifying 1 replica, although replication is not required for DoltLab.

If you want the ability to clone this Hosted Dolt instance, check the box "Enable Dolt Credentials". And finally, if you want to use the SQL workbench feature for this hosted instance (which we recommend) you should also check the box "Create database users for the SQL Workbench".

You will see the hourly cost of running the Hosted Dolt instance displayed above the "Create Deployment" button. Click it, and wait for the deployment to reach the "Started" state.

Once the deployment has come up, the deployment page will display the connection information for both the primary host and the replica, and each will be ready to use. Before connecting a DoltLab instance to the primary host, though, there are a few remaining steps to take to ensure the host has the proper state before connecting DoltLab.

First, click the "Configuration" tab and uncheck the box "behavior_disable_multistatements". DoltLab will need to execute multiple statements against this database when it starts up. You can also, optionally, change the log_level to "debug". This log level setting will make sure executed queries appear in the database logs, which is helpful for debugging.

Click "Save Changes".

Next, navigate to the "Workbench" tab and check the box "Enable Writes". This will allow you to execute writes against this instance from the SQL workbench. Click "Update".

Then, with writes enabled, on this same page, click "Create database" to create the database that DoltLab expects, called dolthubapi.

Finally, create the required users and grants that DoltLab requires by connecting to this deployment and running the following statements:

CREATE USER 'dolthubadmin' IDENTIFIED BY '<password>';
CREATE USER 'dolthubapi' IDENTIFIED BY '<password>';
GRANT ALL ON *.* TO 'dolthubadmin';
GRANT ALL ON dolthubapi.* TO 'dolthubapi';

You can do this by running these statements from the Hosted workbench SQL console, or by connecting to the database using the mysql client connection command on the "Connectivity" tab, and executing these statements from the SQL shell.

This instance is now ready for a DoltLab connection.

Edit DoltLab's Docker Compose file

To connect DoltLab to my-doltlab-db-1, ensure that your DoltLab instance is stopped, and remove references to doltlabdb in DoltLab's docker-compose.yaml file.

You can also remove references to doltlabdb-dolt-data, doltlabdb-dolt-root, doltlabdb-dolt-configs, and doltlabdb-dolt-backups from the volumes section, as these were only necessary for DoltLab's default Dolt server.

  # commenting out all references to doltlabdb
  #
  #
  # doltlabdb:
  #  image: public.ecr.aws/dolthub/doltlab/dolt-sql-server:v1.0.2
  #  command:
  #    -l debug
  #  environment:
  #    DOLT_PASSWORD: "${DOLT_PASSWORD}"
  #    DOLTHUBAPI_PASSWORD: "${DOLTHUBAPI_PASSWORD}"
  #  networks:
  #    - default
  #  volumes:
  #    - doltlabdb-dolt-data:/var/lib/dolt
  #    - doltlabdb-dolt-root:/.dolt
  #    - doltlabdb-dolt-configs:/etc/dolt
  #    - doltlabdb-dolt-backups:/backups
  doltlabenvoy:
     image: envoyproxy/envoy-alpine:v1.18-latest
     command:
       -c /envoy.yaml
...
  doltlabui:
    depends_on:
      # - doltlabdb
      - doltlabenvoy
      - doltlabremoteapi
      - doltlabapi
      - doltlabgraphql
      - doltlabfileserviceapi
...
    networks:
      - default
networks:
  default:
    external:
      name: doltlab
volumes:
  # doltlabdb-dolt-data:
  # doltlabdb-dolt-root:
  # doltlabdb-dolt-configs:
  # doltlabdb-dolt-backups:
  doltlab-remote-storage:
  doltlab-user-uploads:

There's one additional edit to the docker-compose.yaml file to make before we can start DoltLab. Edit the value of the -doltHostargument in the doltlabapi.command section to match the host of the primary my-doltlab-db-1 host. In our example, this would be dolthub-my-doltlab-db-1.dbs.hosted.doltdb.com.

...
  doltlabapi:
...
    command:
      -doltlab
      -outboundInternalServiceEndpointHost doltlabenvoy
      -iterTokenEncKeysFile /iter_token.keys
      -iterTokenDecKeysFile /iter_token.keys
      -doltUser dolthubapi
      -doltHost dolthub-my-doltlab-db-1.dbs.hosted.doltdb.com # update the host to point to the primary deployment
      -doltPort 3306
      -tlsSkipVerify # hosted dolt requires TLS, but we will skip verification for now
...

You will also need to add the argument -tlsSkipVerify to the doltlabapi.command section. Save these changes to the file, and you can now start DoltLab using the start-doltlab.sh script.

Make sure that the DOLT_PASSWORD environment variable matches the password you used when creating user dolthubadmin, and DOLTHUBAPI_PASSWORD matches the password you used when creating user dolthubapi.

Once DoltLab is running successfully against my-doltlab-db-1, you can create a database on DoltLab, for example called test-db, and you will see live changes to the database reflected in the Hosted Dolt workbench:

Serve DoltLab over HTTPS with a TLS reverse proxy

Starting with DoltLab v1.0.5, it is possible to serve a DoltLab instance behind a TLS reverse proxy. You may want to do this if you want to serve your DoltLab instance over HTTPS instead of HTTP. Let's walkthrough an example of how to run a DoltLab instance behind an nginx TLS proxy, running on the same host. We will use doltlab.dolthub.com as our example.

Before you begin you will need to create valid TLS certificates on the DoltLab host. You can provision these from a Certificate Authority or do so with a free tool like certbot. For this example we have created valid certs with certbot, /etc/letsencrypt/live/doltlab.dolthub.com/fullchain.pem and /etc/letsencrypt/live/doltlab.dolthub.com/privkey.pem.

For DoltLab < v2.1.0, shut down your DoltLab instance if it is currently running, then open four new ports on your DoltLab host. These ports will be used to forward requests to DoltLab's existing HTTP ports.

In our example we will using the following new ports: 443, 143, 5443, and 50043. 443 will route requests to port 80 where DoltLab's UI is served. 143 will forward requests to port 100 where DoltLab serves database data from. 5443 will forward requests to 4321, which DoltLab uses to enable user file uploads. And, finally, 50043 will map to 50051, the port used by clients for cloning, pushing, pulling, and fetching data.

At this time you can also close ports 80, 100, 4321, and 50051 on the DoltLab host, as these no longer need to be reachable on the public internet.

Next, edit DoltLab's docker-compose.yaml and amend the arguments under doltlabremoteapi.command:

  doltlabremoteapi:
...
    command:
      # change value of `-http-host`, adding the HOST_IP before the colon and the new TLS database data serving port (143) after the colon
      -http-host "doltlab.dolthub.com:143"
      -http-scheme "https" # add `-http-scheme` with value `https`
      -backingStoreHostNameOverrideKey ":143" # update this with TLS database data serving port as well

There are 2 arguments that need to change, and one new argument to add. As you can see from the above snippet, -http-host needs to be updated with the HOST_IP value followed by a colon and the new TLS database data serving port. Since we use HOST_IP=doltlab.dolthub.com, and we are mapping TLS port 143 to HTTP port 100, our new value is doltlab.dolthub.com:143.

We also need to change the port in the value of -backingStoreHostNameOverrideKey to reflect this same TLS port, so our value here is :143.

Lastly, we need to add the argument -http-scheme https to this command block.

Next, edit the doltlabapi.command section:

  doltlabapi:
...
    command:
      # change the url scheme to `https`
      -websiteURL "https://doltlab.dolthub.com"
      # change the scheme to `https` and the port to the new TLS file upload port (5443)
      -userImportUploadsFileServiceBrowserHost "https://doltlab.dolthub.com:5443"

Here we change the value of -websiteURL to have an https scheme. We also change the value of -userImportUploadsFileServiceBrowserHost so that the url scheme is https and the port after the colon is 5443, the new TLS port we are using to forward requests to DoltLab's user file upload service running on HTTP port 4321.

Save the changes to this file.

Next, install nginx v1.13.10 or higher on the DoltLab host. For this example, the open source version can be installed on Ubuntu with:

sudo apt update
sudo apt install nginx

Ensure nginx is running with:

sudo systemctl status nginx
● nginx.service - nginx - high performance web server
     Loaded: loaded (/lib/systemd/system/nginx.service; enabled; vendor preset: enabled)
     Active: active (running) since Fri 2023-08-18 20:51:50 UTC; 3 days ago
       Docs: https://nginx.org/en/docs/
   Main PID: 2913065 (nginx)
      Tasks: 5 (limit: 18734)
     Memory: 9.8M
     CGroup: /system.slice/nginx.service
             ├─2913065 nginx: master process /usr/sbin/nginx -c /etc/nginx/nginx.conf
             ├─3844184 nginx: worker process
             ├─3844185 nginx: worker process
             ├─3844186 nginx: worker process
             └─3844187 nginx: worker process

Aug 18 20:51:50 ip-10-2-3-125 systemd[1]: Starting nginx - high performance web server...
Aug 18 20:51:50 ip-10-2-3-125 systemd[1]: Started nginx - high performance web server.

Then, edit the nginx configuration file located at /etc/nginx/nginx.conf to be the following:

user  nginx;
worker_processes  auto;

error_log  /var/log/nginx/error.log notice;
pid        /var/run/nginx.pid;


events {
    worker_connections  1024;
}


http {
    include       /etc/nginx/mime.types;
    default_type  application/octet-stream;

    log_format  main  '$remote_addr - $remote_user [$time_local] "$request" '
                      '$status $body_bytes_sent "$http_referer" '
                      '"$http_user_agent" "$http_x_forwarded_for"';

    access_log  /var/log/nginx/access.log  main;

    sendfile        on;
    #tcp_nopush     on;

    keepalive_timeout  65;

    #gzip  on;

    # include /etc/nginx/conf.d/*.conf;

    client_max_body_size 1024M;

    ssl_session_cache   shared:SSL:10m;
    ssl_session_timeout 10m;

    server {
        listen 443 ssl;

        ssl_certificate     /etc/letsencrypt/live/doltlab.dolthub.com/fullchain.pem;
        ssl_certificate_key /etc/letsencrypt/live/doltlab.dolthub.com/privkey.pem;
        ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
        ssl_ciphers         HIGH:!aNULL:!MD5;

        location / {
            proxy_pass http://127.0.0.1:80;
            proxy_http_version 1.1;
            proxy_set_header Host $host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-Host $host;
            proxy_set_header X-Forwarded-Port $server_port;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header X-Forwarded-Proto https;
        }
    }

    server {
        listen 143 ssl;

        ssl_certificate     /etc/letsencrypt/live/doltlab.dolthub.com/fullchain.pem;
        ssl_certificate_key /etc/letsencrypt/live/doltlab.dolthub.com/privkey.pem;
        ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
        ssl_ciphers         HIGH:!aNULL:!MD5;

        location / {
            proxy_pass http://127.0.0.1:100;
            proxy_http_version 1.1;
            proxy_set_header Host $host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-Host $host;
            proxy_set_header X-Forwarded-Port $server_port;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header X-Forwarded-Proto https;
        }
    }

    server {
        listen 5443 ssl;

        ssl_certificate     /etc/letsencrypt/live/doltlab.dolthub.com/fullchain.pem;
        ssl_certificate_key /etc/letsencrypt/live/doltlab.dolthub.com/privkey.pem;
        ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
        ssl_ciphers         HIGH:!aNULL:!MD5;

        location / {
            proxy_pass http://127.0.0.1:4321;
            proxy_http_version 1.1;
            proxy_set_header Host $host;
            proxy_set_header X-Real-IP $remote_addr;
            proxy_set_header X-Forwarded-Host $host;
            proxy_set_header X-Forwarded-Port $server_port;
            proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
            proxy_set_header X-Forwarded-Proto https;
        }
    }

    server {
        listen 50043 ssl http2;

        ssl_certificate     /etc/letsencrypt/live/doltlab.dolthub.com/fullchain.pem;
        ssl_certificate_key /etc/letsencrypt/live/doltlab.dolthub.com/privkey.pem;
        ssl_protocols       TLSv1 TLSv1.1 TLSv1.2;
        ssl_ciphers         HIGH:!aNULL:!MD5;

        location / {
            grpc_pass grpc://127.0.0.1:50051;
        }
    }
}

The above configuration file includes server blocks that route the new TLS ports to the proper HTTP ports of DoltLab and use the TLS certificates we created earlier with certbot. Importantly, the server block for 50043 used by clients for cloning, pushing, pulling, and fetching must be configured with http2 and as a grpc_pass.

Save the changes to the configuration file and reload nginx with: nginx -s reload. This will make the configuration changes take effect.

Finally, restart DoltLab using the ./start-doltlab.sh script with two additional environment variables:

export USE_HTTPS=1 # enable https support in the DoltLab frontend
export DOLTLAB_REMOTE_PORT=50043 # set to new TLS port for cloning/pushing/pulling/fetching (50043)
...
./start-doltlab.sh

Once your DoltLab instance comes up, it will be served on HTTPS via the nginx TLS proxy.

Serve DoltLab over HTTPS natively

Starting with DoltLab v1.0.6, it is possible to run DoltLab over HTTPS with TLS natively. To do so, make sure that port 443 is open on the host running DoltLab (as well as the other required ports 100, 4321, and 50051) and that you have a valid TLS certificate that uses the HOST_IP of the DoltLab host. We recommend creating a TLS certificate using certbot.

For DoltLab < v2.1.0, to start DoltLab with TLS, you will run the ./start-doltlab.sh script with the argument https, and will need to supply two additional environment variables:

...
export TLS_CERT_CHAIN=/path/to/tls/certificate/chain
export TLS_PRIVATE_KEY=/path/to/tls/private/key
./start-doltlab.sh https

Once the services are spun up, DoltLab will be available at https://${HOST_IP}.

Improve DoltLab Performance

Starting with DoltLab v1.1.0, it is possible to limit the number of concurrent Jobs running on a DoltLab host by adding optional arguments to the doltlabapi block of the docker-compose.yaml or docker-compose-tls.yaml files.

When user's upload files on a DoltLab instance, or merge a pull request, DoltLab creates a Job corresponding to this work. These Jobs spawn new Docker containers that performs the required work.

By default, DoltLab imposes no limit to the number of concurrent Jobs that can be spawned. As a result, a DoltLab host might experience resources exhaustion as the Docker engine uses all available host resources for managing it's containers.

For DoltLab < v2.1.0 and >= v1.1.0, to prevent resource exhaustion, the following can be added to limit the number of concurrent Jobs, ensuring DoltLab will not run more jobs than the configured limit, at any one time:

# docker-compose.yaml
---
doltlabapi:
---
command:
---
-jobConcurrencyLimit "5"
-jobConcurrencyLoopSeconds "10"
-jobMaxRetries "5"

-jobConcurrencyLimit limits number of concurrent Jobs a DoltLab instance will run at any given time. A value of 0 indicates no limit.

-jobConcurrencyLoopSeconds is the number of seconds Job Scheduler will wait before looking for more Jobs to schedule. Default is 10 seconds.

jobMaxRetries is the number of times the Job Scheduler will retry scheduling a Job before permanently giving up, requiring the Job to be recreated.

Serve DoltLab behind an AWS Network Load Balancer

The following section describes how to setup an AWS Network Load Balancer (NLB) for a DoltLab instance. This guide will be using DoltLab v2.0.8.

First, setup DoltLab v2.0.8 on an AWS EC2 host in the same Virtual Private Cloud (VPC) where your NLB will run.

If this instance should only be accessible by the NLB, ensure that the DoltLab host is created in a private subnet and does not have public IP address.

After setting up your DoltLab host, edit the host's inbound security group rules to allow all traffic on ports: 80/443, 100, 4321, 50051, and 2001.

Because the host is in a private subnet with no public IP though, only the NLB will be able to connect to the host on these ports.

Next, edit the envoy.tmpl file included with DoltLab so that it includes the following listener for port 2001:

  - name: health_check_listener
    address:
      socket_address: { address: 0.0.0.0, port_value: 2001 }
    traffic_direction: inbound
    filter_chains:
      - filters:
        - name: envoy.filters.network.http_connection_manager
          typed_config:
            "@type": type.googleapis.com/envoy.extensions.filters.network.http_connection_manager.v3.HttpConnectionManager
            codec_type: auto
            stat_prefix: http
            route_config:
              virtual_hosts:
              - name: http
                domains: ["*"]
                routes:
                - match: { path: "/" }
                  direct_response:
                    status: "200"
                    body:
                      inline_string: "live\n"
            http_filters:
            - name: envoy.filters.http.router
            access_log:
            - name: envoy.access_loggers.stream
              typed_config:
                "@type": type.googleapis.com/envoy.extensions.access_loggers.stream.v3.StdoutAccessLog
                log_format:
                  text_format_source:
                      [%START_TIME%]
                      "%REQ(:METHOD)% %REQ(X-ENVOY-ORIGINAL-PATH?:PATH)% %PROTOCOL%"
                      %RESPONSE_CODE% %GRPC_STATUS% %RESPONSE_FLAGS%
                      %BYTES_RECEIVED% %BYTES_SENT%
                      %DURATION% %RESP(X-ENVOY-UPSTREAM-SERVICE-TIME)%
                      "%DOWNSTREAM_REMOTE_ADDRESS_WITHOUT_PORT%"
                      "%REQ(USER-AGENT)%" "%REQ(X-REQUEST-ID)%" "%REQ(:AUTHORITY)%"
                      "%UPSTREAM_HOST%"
                      "%UPSTREAM_TRANSPORT_FAILURE_REASON%"

This listener will allow AWS to perform health checks against port 2001 during the NLB creation process.

Similarly, edit the docker-compose.yaml file to expose port 2001 in the doltlabenvoy block:

  doltlabenvoy:
     ...
     ports:
      ...
       - "2001:2001"

Next, in AWS, create target groups for each DoltLab port that the NLB will forward requests to. These ports are: 80/443, 100, 4321, and 50051.

When creating the target groups, select Instances as the target type. Then, select TCP as the port protocol, followed by the port to use for the target group. In this example we will map all target group ports to their corresponding DoltLab port, ie 80:80, 100:100, 4321:4321 and 50051:50051. Select the same VPC used by your DoltLab host as well.

During target group creation, in the Health Checks section, click Advanced health check settings and select Override to specify the port to perform health checks on. Here, enter 2001, the port we added to DoltLab's envoy configuration file. We will use this same port for all target group health checks.

After clicking Next, you will register targets for your new target group. Here you should see your DoltLab host. Select it and specify the port the target group will forward to.

Click Include as pending below, then click Create target group.

Once you've created your target groups you can create the NLB.

Be sure to select the Network Load balancer as the other types of load balancers may require different configurations.

Then, create an NLB in the same VPC and subnet as your DoltLab host that uses Scheme: Internet-facing and Ip address type: IPV4.

Additionally, select the the same availability zone that your DoltLab host uses. You can use the default security group for your NLB, however the ingress rules for this group will need to be updated before inbound traffic will be able to reach your NLB.

In the Listeners section, add listeners for each target group you created, specifying the NLB port to use for each one. But again, in this example we will forward on the same port. Click Create load balancer.

It make take a few minutes for the NLB to become ready. After it does, check each target group you created and ensure they are all healthy.

Next, edit the inbound rules for the security group attached to the NLB you created so that it allows connections on the listening ports.

On the NLB page you should now see the DNS name of your NLB which can be used to connect to your DoltLab instance.

Restart your DoltLab instance supplying this DNS name as the HOST_IP, and your DoltLab instance will now be running exclusively through the NLB.