Cluster overview¶

On the Cluster Overview you can list clusters, get more information, and manage them.

Filter and pagination¶

You can filter all clusters:

• By checked status
• By name of the cluster owner
• By project name
• By Elixir ID of the cluster owner
• By cluster name

Click Filter.
You can set how many clusters you want to see on a page and scroll through the pages.

Cluster information card¶

See the name, the status, the project, and the creator of the cluster.
Click on the name to get to the Detail page. Click on the project name to get to the Project overview.

How to connect¶

Click How to connect. A list with connection information opens depending on your cluster configuration.

Scale-Up your cluster¶

To expand your cluster and add more workers, click Scale-up.

A modal opens where you can specify the batch you want to expand or add a new worker batch. Click Add new Worker Batch to add a new worker batch where you can specify the flavor, and the worker number.

This process only starts the worker nodes. You have to configure your cluster before your cluster can use the new worker nodes.

Configure your cluster¶

Your have to reconfigure your master node to use the new worker nodes. A modal opens after starting your new worker nodes, which shows you the necessary steps.

Connect to your master instance with SSH.
Download the scaling script:

wget -O scaling.py https://raw.githubusercontent.com/deNBI/user_scripts/master/bibigrid/scaling.py

Run the scaling script:

python3 scaling.py 

The scaling script requests a password. If you already closed the modal or have no access to the password shown in the modal, you can generate a new password.
Click Generate Password at the bottom of the cluster information card. A modal opens with a new generated password.

You can use the new workers after you ran the script.
You can test with the command sinfo if the script added your workers correctly.

If the new worker is still missing, you can try to restart slurm:

sudo /etc/init.d/slurmctld restart

Scale-Down your cluster¶

To reduce your cluster and remove workers, click Scale-Down.

A modal opens where you can select the batches you want to reduce, and the number of workers you want to remove.

This process only removes the worker nodes. You have to configure your cluster before your cluster recognizes the removed worker nodes.

Configure your cluster¶

Your have to reconfigure your master node to recognize the removed worker nodes. A modal opens after removing the worker nodes, which shows you the necessary steps.

Connect to your master instance with SSH.
Download the scaling script:

wget -O scaling.py https://raw.githubusercontent.com/deNBI/user_scripts/master/bibigrid/scaling.py

Run the scaling script:

python3 scaling.py 

The scaling script requests a password. If you already closed the modal or have no access to the password shown in the modal, you can generate a new password.
Click Generate Password at the bottom of the cluster information card. A modal opens with a new generated password.

You can test with the command sinfo if the script removed your workers correctly.

If your workers aren't removed, you can try to restart slurm:

sudo /etc/init.d/slurmctld restart

Stop your cluster¶

To stop your cluster, click Stop. A modal opens, asking you to confirm your action.

Delete your cluster¶

To delete your cluster, click Delete. A modal opens, asking you to confirm your action.

If you have volumes attached to a node of your cluster, they get detached but not deleted.