You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
With the new async interface #13 Different containers can be handle (as long as different ports are requested). Containers are only shutdown if not jobs are running at all. This allows to have multiple containers and route the requests in a client-server interface using the unique metadata of a given job to properly handle the communication.
That sounds like it would work. If it helps, you can actually shut down a Cerise container while jobs are running. You just have to make sure to not delete it (a new container wouldn't know about existing jobs) and to start it up again before submitting more jobs, or requesting the status of runnning jobs (otherwise you'll get no response from the REST API, since the server behind it is not running).
A stopped Cerise container doesn't do any staging though, so keeping them running means that result downloads will start as soon as the results are available. That will probably make our users happy.
The lie_md service should keep multiple Cerise dockers running, and keep track of which users owns which container, and route requests appropriately.
The text was updated successfully, but these errors were encountered: