Overview: Model II
A CLC Server with grid integration allows jobs to be offloaded from a master server to grid nodes using local, third party, job scheduling software. When a job is started by the scheduler on a grid node in the local cluster, a stand-alone executable referred to as a CLC Grid Worker is started, and the CLC job sent by the user is executed.Grid nodes must have enough memory and space to carry out the task(s) submitted to them. When a single CLC tool is run, that job runs in its entirety on a single grid node. How tasks within a workflow are executed (e.g. on a single node, or split across nodes) is configurable (see Workflow distribution options).
Two locations beyond those needed for a single CLC Server setup need to be specified in grid setups. Both are configured within grid presets (see Configuring grid presets). They are:
- A CLC Grid Worker location A CLC Grid Worker contains a JRE matching the master CLC Server JRE, as well as the plugins installed on the master, associated settings files, libraries, etc. These are extracted from the installation area of the CLC Server software and deployed to the location specified in a grid preset when the grid preset is saved.
- A shared work directory This directory is used for communication between the CLC Server and a CLC Grid Worker. Each grid job gets its own, temporary, sub-directory under this area where temporary files are written.