November 29, 2019
Is Columbus+ able to address more than one cluster instance?
Columbus+ uses two files (located in /etc/columbus) to store the parameters of a given cluster; a config file (*.config) which stores cluster connection parameters, and a shell script (*.sh) which stores the job submission parameters.
By using multiple config files (and associated shell scrips), where each configuration corresponds to a different cluster instance, its possible to address multiple clusters. Each unique config file will be represented by an entry in the "Run Analysis on" drop-down menu of the Batch Analysis view, meaning users can choose which cluster to use when submitting their jobs. This is also the only way to configure different "queues" on the same cluster.
There is one major constaint with the current framework; both input_path (where input image data is stored) and output_path (where output results files are stored) must be identical between all configurations.
Contact the Informatics Support team (firstname.lastname@example.org) for further information.