Using the server

Launching the server

On the first run, you can either manually create a 'server.config' file or run:

$ python server.py --setup

This will execute the auto configuration of the server, creating a list of TF compatible GPU and CPU compute devices. The server can then be started using the following command:

$ python server.py

The following (optional) flags can be used to specify how the server instance is configured.

usage: server.py [-h] [--jobdir JOBDIR] [--logdir LOGDIR]
                 [--gpus [{0,1,2,3,4,5,6,7} [{0,1,2,3,4,5,6,7} ...]]]
                 [--local] [--setup] [--use_config USE_CONFIG]

Sequitr server process

optional arguments:
  -h, --help            show this help message and exit
  --jobdir JOBDIR       Path to job directory
  --logdir LOGDIR       Path to log directory
  --gpus [{0,1,2,3,4,5,6,7} [{0,1,2,3,4,5,6,7} ...]]
                        Specify the gpus which can be used for processing
  --local               Running a local server only. Prevents pinging.
  --setup               Runs the setup configuration to determine hardware
                        specs, and generate the config file. On server
                        restart, the config will persist.
  --use_config USE_CONFIG
                        Use a specific (non-default) server config file. The
                        default option, without specifying the --config flag
                        is server.config. This is the file written out when
                        the --setup flag has previously been used to create a
                        server configuation.