Backend is node app connected with priviled access to the contracts with ethers.js and communicates with the frontend clients with socket.io events and also exposes simple REST api.
Install dependencies
nvm use
npm install
Start local database server in Docker
npm run db
Start local backend connected to the local contracts
npm run dev
This will copy ../webapp/contracts/development.json and launch nodemon - if contracts are redeployed,
backend has to be restarted or new contracts info copied manually.
To be able to run staging or production contracts .env file has to be filled with corresponding mnemonic and AWS vars.
Start staging backend connected to the staging contracts
npm run dev:staging
Start production backend connected to the production contracts
npm run dev:prod
Backend startup can take long time depending on the environment chosen and on how synced is local database.
Backend is synced and ready when
https://localhost:3399 returns ethernalCache: 'ok'
Backend unit test can be run by
npm test
Prerequisites: heroku cli, docker
You have to be logged in in heroku cli first.
Deploy staging environment (https://ethernal-be-alpha.herokuapp.com)
npm run deploy:staging
Deploy production environment (https://ethernal-cache-1.herokuapp.com)
npm run deploy:prod1
Deploy alternative production environment (https://ethernal-cache-2.herokuapp.com)
npm run deploy:prod2
Both production backend are connected to the same database, to which scheme they write is specified by SCHEMA_PREFIX
environment variable.
This way we can have one backend using one scheme and other one indexing another from scratch. Reindexing database takes long time so it is important to reindex always in advance and only if really necessary.
When second scheme is synced, default scheme can be replaced like so:
ALTER SCHEMA primary RENAME TO backup
ALTER SCHEMA secondary RENAME TO primary;
This change can be done live with primary and secondary backend running, but if for some reason would be necessary to switch back to backup, the primary backend has to be stopped before switch. Also primary backend has to be switched if the secondary backend is not running during the switch. If the redeploy of the primary backend is necessary switch should be executed during the downtime.
To force local reindex of database, the scheme can be just deleted or the whole database cleared by running:
npm run db:reset
Admin wallet can be rotated by specifying both MNEMONIC and OLD_MNEMONIC (current admin) environment variables.