Docker based Ergo setup. Somewhat similar to Ergo Bootstrap except it offers much less options and is not NixOS-based.
Install Docker, then follow the instructions below for the components you'd like to run.
If Docker doesn't come with the docker compose command on your system, install Docker Compose separately. Once installed, you'll either have docker compose or docker-compose available.
The docs below use the
docker composecommand which may have to be replaced bydocker-composeon some systems (e.g. Ubuntu 20 and lower).
Containers are configured to expose certain ports for convenient access within a home lab context. Modify the
portsvalues in thedocker-compose.ymlfiles if needed.
# Volume and network expected by node compose file.
docker volume create ergo_node
docker network create ergo-node
# Starting the service
cd node
docker compose up -d
# Stopping the service (still from within the node directory)
# Using stop before/instead of down seems to cause less db corruption issues
docker compose stop node
docker compose down
Edit build.sh (or build.bat if using Windows) and set the EXPLORER_VERSION variable to the desired Explorer version. You can use any tag from the explorer repository: https://github.com/ergoplatform/explorer-backend.
If using another node than the one defined in this stack, edit the
master-nodesfield inexplorer\explorer-backend.confto point it to your node.
# Run the build script
cd explorer
./build.sh
# Choose a password for the database
# Do this *before* starting the db service for the first time.
# If wanting to change the db password later, change it through a db client,
# then update it in db/db.secret.
echo POSTGRES_PASSWORD=some-pw-you-can-choose > db/db.secret
# Same password but different env variable for the GraphQL service
echo DB_USER_PWD=some-pw-you-can-choose >> db/db.secret
# Create a named volume for Redis
docker volume create ergo_redis
# Start all services in one go...
docker compose up -d --build
# ...or only the ones you need
docker compose build db grabber api
docker compose up --no-start
docker compose start db grabber
docker compose start api
# Check their status
docker ps --filter name=explorer -a
# Stop all explorer services
docker compose down
# ...or
docker compose stop api grabber
docker compose stop db
Before building the ui service, edit the API arg of the ui image in docker-compose.yml to point to the (external) URL of your api service.
The graphql server will run over http on port 3001. To use it with clients requiring https, put it behind a reverse proxy.
Unlike the node, the database volume is mapped, not named. Mapped volumes make it easier to handle Postresql upgrades.
This means you may have to edit the path in explorer/docker-compose.yml and ensure the specified path exists on your system:
services:
db:
...
volumes:
# Mapped volume for easier Postgres upgrades.
# Make sure the path exists or edit it here.
# See also readme_pg_upgrade.md
- /var/lib/explorer_pg/14/data:/var/lib/postgresql/data
...
Syncing from scratch will take a long time (weeks). Two solutions:
1. Drop indexes and constraints from the database and restore them once initial sync is done.
2. Get a snapshot from someone
Sync performance seems to have improved significantly. There have been several reports of full syncs in 48h or less. Still, if in a hurry, above points remain valid.
If wanting to backup your db, migrate it or provide a dump for someone else, we recommend the following:
docker exec -it explorer-db-1 pg_dump -Fc -v --schema=public --file /var/lib/postgresql/data/explorer.dmp --username=ergo ergo
This will use the pg_dump present within the db container and save the dump to the mapped db volume. Note the Fc flag which outputs the dump in custom format resulting in smaller dump size and faster processing.
To restore, start the db service only at first, no grabber, no utx-tracker. Copy the dump file to the mapped db volume so that the db container can see it. Then run:
docker exec -it explorer-db-1 pg_restore -v --clean --create --dbname=postgres --username=ergo /var/lib/postgresql/data/explorer.dmp
This will use the pg_restore present in the db container and read the dump from the mapped volume.
Alternatively, if you have pg_restore or pgAdmin on the host or a remote and can connect to the db from the host/remote, you can load the dump from there. Just make sure your pg_restore's version is not behind the postgresql version running on the db container.