Read Me

pmHPC - Poor Man's HPC


INSTALLATION // CONFIGURATION // USAGE:


1) Deploy pmhpc_srv anywhere on master node.

2) Deploy pmhpc_cl anywhere to all client (processing) nodes

3) Default settings in the client script are:
   - local bind address:port -- localhost:8001
   - server address:port -- localhost:9001
   You probably want to change those to something different

4) Default settings on job manager (master node):
   - local bind address:port -- localhost:9001
   Same as with the client, change to something else, 
   make sure the address matches client configuration (see step 3)

5) Add all clients to the configuration file config.xml

6) Start client daemon on all client (processing) nodes

7) Deploy code with './pmhpc_tool.py deploy'

8) Start job manager process on master node './job_manager.py'

9) Initiate all jobs './pmhpc_tool.py initalljobs'
   This (most likely) tell you that all jobs initiated and status is 'OK'.
   That only means that jobs were started, they might be still running.