I ran an ML job on the notebook cluster which was running fine but overnight. When I came back in the morning the cluster had shutdown and my job was killed. Is that the right behavior?
@ilya_at_vivid You can use the Swagger page to create a new cluster with the autopause option set to false. You will need to sign in before being able to run anything.
@ilya_at_vivid If you start a Terra notebook cluster, it will go into auto-pause mode after 30 minutes of inactivity. The inactivity means that you are not using the browser for more than 30 minutes. However, the cluster should not be deleted. You will be able to hit the PLAY button again to re-start the cluster. The cluster is only deleted by explicitly hitting the TRASH icon to delete the cluster.
Thanks, I understand. It looks like I can control the auto-pause timing. Do you have an example of how that can be done?
@ilya_at_vivid Additionally, if you have questions that are specific to the Terra platform, can you please post them to our new Terra Community Forum? As we transition to the new platform, our efforts to answer forum questions will also be focused there! Additionally, you can use the Contact Us widget within the Terra platform! Here is a screenshot of the sidebar menu to access the Contact Us as well as a direct link to the Community Forum (plus documentation!):
Cool, I'll give it a try. Is there a phython sdk/library?
The default autopause is set to 30 minutes, so if your computer sleeps the cluster will be paused (and your job killed). If you have a job you'd like to run overnight, you can stop your computer from sleeping, or create a cluster through swagger and set the autopause to false to turn that feature off. The swagger API information can be found here: https://notebooks.firecloud.org/#!/cluster/createCluster