Hello everybody,
I am looking for some help to deploy a local galaxy instance which will be using a cluster for execution. I am trying to create the best file system schema possible so I only have to share the minimum number of directories between cluster nodes and galaxy server. Galaxy documentation (https://wiki.galaxyproject.org/Admin/Config/Performance/Cluster) currently state that: The path to Galaxy must be exactly the same on both the nodes and the application server. I know that some directories are needed because, for example, galaxy tools are referenced using exact path but my idea is to separate the shared parts from the ones that are only accessible from the galaxy server. Something like: galaxy/src: config files, .loc files, galaxy/shared: includes /database (including /files, /drmaa,/job_workingdirectory), /tools, /tool-data, /requirements (dependencies)
Would you recommend trying to separate this ? I think the shared FS requirement will be modified in the near future. I dont have very clear how galaxy creates the .sh files that will be executed but, is it possible that these reference any other path besides those I have indicated as shared? does the whole galaxy base-path needs to be shared? if so, what for? for those that have an instance using a cluster, what kind of file system organization do you use?
Thanks very much for your help.