tractor-engine
The engine maintains the central job queue and assigns tasks to servers on the farm. It is also the central clearinghouse for status information reported from the blades. The engine also delivers the web content required by the Tractor Dashboard.Usage:tractor-engine [options]
Options:
Option | Description |
---|---|
--version | Print version information, and exit. |
-h, --help | Print usage summary, and exit. |
--configdir=path or -c path | Specify the directory containing tractor configuration files, default is /var/spool/tractor/config. Tractor-engine also checks for the environment variable TRACTOR_CONFIG as well. |
--log=filename | Specifies the 'filename' used for logging the engine's own diagnostics. Default is stderr. |
--verbose or -v | Increase logging level to from NOTICE to INFO. |
--quiet or -q | Decrease logging level from NOTICE to SEVERE. |
--debug | Increase logging level to DEBUG. |
--trace | Increase logging level to TRACE. |
--httpdebug | Log inbound http requests |
--nshippers=n or -n nnn | Set the number of thread created to handle i/o (default=ncores). |
--maxdispatch=n or -D nnn | Restrict concurrent dispatches (and license use). |
--port=n | Specify the engine's inbound connection port (default=80). |
--port=addr:port | Restrict the inbound listener port to the network interface given by address - using port number port. By default the engine will listen for connections on port 80, on all interfaces. Example usage: --port=8080 or --port=127.0.0.1:9999 |
--supersede | Allows a newly started engine process to gracefully take over from a still-running previous engine. The new engine sends a "drain and exit" message to the old engine, then waits for the old one to exit before proceeding to become the new engine service. In drain mode, the old engine stops assigning new work to requesting blades, and then exits when all previously launched tasks are complete. Partially complete jobs will remain in the queue and will continue being processed when the new engine takes over. |
--paused | Causes a newly restarted engine to immediately enter the "pause all dispatching" state. Hence no new commands will be assigned to blades until an administrator manually re-enables dispatching using the red "play" button on the Dashboard's "Gears->Admin" tab. Even when the engine restarts in --paused mode, any existing commands that are marked as still-active in the job database will still be considered active by the new engine. In those cases, the still-running blades and the new engine will communicate to sort out which of those commands are actually still running. |
--startupdelay=n | Engine simply sleeps for n seconds at startup. |
--daemon | Engine auto-backgrounds in system service mode. |
--pidfile=filename | In daemon mode, write final PID to filename. |
--sigdfl=1 | Do not override default SEGV/BUS handlers. |