This blog post will go over the procedure for migrating a Tator deployment based on Kubernetes (pre-1.1.0) to Docker Compose (1.1.0 and later). Because all dependencies use the same Docker images as version 1.0.0, this migration procedure is fairly simple and consists of simply pointing the new compose deployment to the directory that was previously in use.
Set the DATA_DIR variable in the .env file
Open the .env file in the root directory of Tator and edit the DATA_DIR to point to the local directory that was being used by the Kubernetes deployment. If you used the install.sh script to install Tator, you would change:
DATA_DIR=${HOME}/tator_data
to
DATA_DIR=/media/kubernetes_share
Inspect other environment variables
The .env file contains the same default values found in values.yaml from the Kubernetes-based deployment. If you used the install script and never modified values.yaml, no modifications to .env should be needed. Inspect .env and change any values to correspond to values.yaml (for example, Postgresql username/password, Minio credentials, Django secret key, etc.)
Install Tator
Follow the normal procedure for installing Tator on compose, by running:
make tator
All your previously used data should still be present.