This upgrade guide is only for those who have installed OpenPAI >= v1.0.0
, and want to upgrade current cluster to a newer version. For example, upgrade from v1.0.0
to v1.1.0
. If you want to upgrade older version before v1.0.0
to version >= v1.0.0
. Please refer to Installation Guide.
The upgrade process is mainly about modifying services-configuration.yaml
and using paictl
. If you are not familiar about them, please first refer to here to set up paictl
and services-configuration.yaml
.
Notice: We recommend those who are upgrading from version < v1.6.0
(not including v1.6.0
) to check this issue before follow the steps below.
First, launch a dev box container of the current PAI version, stop all services by:
./paictl.py service stop
The command will ask you for the cluster id for confirmation. If you forget it, another command ./paictl.py config get-id
will help you.
If you don't want to affect current running jobs, please skip stopping storage-manager
by using the following command:
./paictl.py service stop --skip-service-list storage-manager
Use exit
to leave the dev box container. And remove it by:
sudo docker stop dev-box
sudo docker rm dev-box
Now, launch a dev box container of the new version. For example, if you want to upgrade to v1.1.0
, you should use docker openpai/dev-box:v1.1.0
.
Then, retrieve your configuration by:
./paictl.py config pull -o <config-folder>
Find the following section in <config-folder>/services-configuration.yaml
:
cluster:
# the docker registry to store docker images that contain system services like Frameworklauncher, Hadoop, etc.
docker-registry:
......
tag: v1.0.0
......
Change the tag
to the version you want to upgrade to, e.g. v1.1.0
, then save the file.
Push the modifed services-configuration.yaml
by:
./paictl.py config push -p <config-folder> -m service
Start all PAI services by:
./paictl.py service start
If you didn't stop storage-manager
, start other services by:
./paictl.py service start --skip-service-list storage-manager
After all the services are started, your OpenPAI cluster is successfully upgraded.