Self-hosting Best Practices
This document is intended to share the best practices to follow when running Appsmith in production.
Choice of deployment platform
We recommend running Appsmith using Kubernetes in production. Using Kubernetes ensures your deployment will scale well. You will also not need to host an external MongoDB or Redis instance for high availability as our helm chart will run them as a separate pod for you.
If you decide to use a serverless deployment platform like ECS, it will require you to setup an external MongoDB and Redis instance. We recommend using MongoDB Atlas and Elasticache.
- Setup a separate instance for Production & Staging. You may also choose to have one for development.
- Enable scheduled backups to run nightly on all your instances.
- Have your backups synced to S3
- Add a monitoring integration like BetterUptime or UptimeRobot, to monitor the uptime of your production Appsmith instance.
- Always ensure you have an NFS persistant volume of 3GB. On AWS, EFS is what we recommend.
- If using ECS, enable CloudWatch logging, so that you can easily retrieve your Appsmith logs.
- Enable HTTPS for your Appsmith instance and deploy it under a subdomain as
t3.medium, or a server of equivalent capacity, is the minimum requirement for running Appsmith. This should scale well for hundreds of users. If you expect ~500 concurrent users, consider picking a
t3.large, or bigger.
- Ensure you have a healthy free disk space, at least 10-15 GB free, on your server at all times.
- Do not run other self-hosted applications on the same server as Appsmith.
- Disable form login & configure a federated login method like Google OAuth or SAML SSO.
- If you are using form login, enable email verification & disable allowing all users to signup. Users can still signup if they are invited.
- Configure an SMTP server as this is necessary for email verification of new signups and resetting account passwords.
- If deployed on AWS EC2, consider changing the Instance Metadata version to IMDsv2, which is more secure, and harder to do SSRF attacks against.
- Take a backup of your Appsmith Development, Staging & Production instances. Backup
- Always perform a manual upgrade of your Staging / Development instance first before you attempt it on the production instance.
- Check if you need to upgrade to a check point release.
- Verify that all your applications are working on the Staging & Development instances in both the edit & view modes.
- Perform the upgrade on the Production instance after taking a backup.
- Verify that all your applications are working on the production instance in both edit and view modes.
- If you notice any degradation in your experience after the upgrade, restore the backup and reach out to our support via the chat widget on this page.