If you have an existing Arvados installation using Docker 1.9 and wish to update to Docker 1.10+, you must migrate the Docker images stored in Keep.
The @arv-migrate-docker19@ tool converts Docker images stored in Arvados from image format v1 (Docker <= 1.9) to image format v2 (Docker >= 1.10).
If you have an existing Arvados installation using Docker 1.9 and wish to update to Docker 1.10+, you must migrate the Docker images stored in Keep.
The @arv-migrate-docker19@ tool converts Docker images stored in Arvados from image format v1 (Docker <= 1.9) to image format v2 (Docker >= 1.10).
Usage:
# Install arvados/migrate-docker19 image: @docker pull arvados/migrate-docker19:1.0@. If you're unable to do this, you can run @arvados/docker/migrate-docker19/build.sh@ to create @arvados/migrate-docker19@ Docker image.
Usage:
# Install arvados/migrate-docker19 image: @docker pull arvados/migrate-docker19:1.0@. If you're unable to do this, you can run @arvados/docker/migrate-docker19/build.sh@ to create @arvados/migrate-docker19@ Docker image.
# Set ARVADOS_API_HOST and ARVADOS_API_TOKEN to the cluster you want to migrate.
# Your temporary directory should have the size of all layers of the biggest image in the cluster, this is hard to estimate, but you can start with five times that size. You can set up a different directory by using the @--tmp-dir@ switch. Make sure that the user running the docker daemon has permissions to write in that directory.
# Run @arv-migrate-docker19 --dry-run@ from the Arvados Python SDK on the host (not in a container). This will print out some information useful for the migration.
# Set ARVADOS_API_HOST and ARVADOS_API_TOKEN to the cluster you want to migrate.
# Your temporary directory should have the size of all layers of the biggest image in the cluster, this is hard to estimate, but you can start with five times that size. You can set up a different directory by using the @--tmp-dir@ switch. Make sure that the user running the docker daemon has permissions to write in that directory.
# Run @arv-migrate-docker19 --dry-run@ from the Arvados Python SDK on the host (not in a container). This will print out some information useful for the migration.