doc/html/launch_plugins.shtml

<!--#include virtual="header.txt"-->

<h1><a name="top">Launch Plugin API</a></h1>

<h2> Overview</h2>
<p> This document describes the launch plugin that is responsible for
  launching a parallel task in SLURM and the API that defines them. It
  is intended as a resource to programmers wishing to write their own
  launch plugin.

<p><span class="commandline">const char
plugin_name[]="<i>launch&nbsp;SLURM&nbsp;plugin</i>"</span>
<p style="margin-left:.2in">

<p><span class="commandline">const char
plugin_type[]="<i>launch/slurm</i>"</span><br>
<p style="margin-left:.2in">

<ul>
<li><b>aprun</b>&#151;Use Cray's aprun command to launch tasks - used on Cray
systems with ALPS installed.</li>
<li><b>poe</b>&#151;Use IBM's poe command to launch tasks - used on systems
  IBM's parallel environment (PE) installed.</li>
<li><b>runjob</b>&#151;Use IBM's runjob command to launch tasks - used on
  BlueGene/Q systems.</li>
<li><b>slurm</b>&#151;Use SLURM's default launching infrastructure</li>
</ul>
<p>The programmer is urged to study
<span class="commandline">src/plugins/launch/slurm/launch_slurm.c</span>
for a sample implementation of a SLURM launch plugin.
<p class="footer"><a href="#top">top</a>

<h2>API Functions</h2>

<p class="commandline"> int init(void)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Called when the plugin is loaded, before any other functions are
  called. Put global initialization here.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> int fini(void)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Called when the plugin is removed. Clear any allocated storage here.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> int launch_p_setup_srun_opt(char **rest)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Sets up the srun operation.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> rest:</span> extra parameters on the
  command line not processed by srun
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> int launch_p_handle_multi_prog_verify(int command_pos)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Is called to verify a multi-prog file if verifying needs to be done.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> command_pos:</span> to be used with
  global opt variable to tell which spot the command is in opt.argv.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">1</span> if handled, or<br>
  <span class="commandline">0</span> if not.

<p class="commandline"> int launch_p_create_job_step(srun_job_t *job,
  bool use_all_cpus, void (*signal_function)(int), sig_atomic_t
  *destroy_job)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Creates the job step.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> job:</span> the job to run.<br>
  <span class="commandline"> use_all_cpus:</span> choice whether to use
  all cpus.<br>
  <span class="commandline"> signal_function:</span> function that
  handles the signals coming in.<br>
  <span class="commandline"> destroy_job:</span> pointer to a global
  flag signifying if the job was canceled while allocating.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> launch_p_step_launch(srun_job_t *job,
  slurm_step_io_fds_t *cio_fds, uint32_t *global_rc)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Launches the job step.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> job:</span> the job to launch.<br>
  <span class="commandline"> cio_fds:</span> filled in io descriptors<br>
  <span class="commandline"> global_rc:</span> srun global return code.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> int launch_p_step_wait(srun_job_t *job, bool
  got_alloc)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Waits for the job to be finished.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> job:</span> the job to wait for.<br>
  <span class="commandline"> got_alloc:</span> if the resource
  allocation was created inside srun.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> int launch_p_step_terminate(void)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Terminates the job step.
<p style="margin-left:.2in"><b>Returns</b>: <br>
  <span class="commandline">SLURM_SUCCESS</span> on success, or<br>
  <span class="commandline">SLURM_ERROR</span> on failure.

<p class="commandline"> void launch_p_print_status(void)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Gets the status of the job.

<p class="commandline"> void launch_p_fwd_signal(int signal)
<p style="margin-left:.2in"><b>Description</b>:<br>
  Sends a forward signal to any underlying tasks.
<p style="margin-left:.2in"><b>Arguments</b>: <br>
  <span class="commandline"> signal:</span> the signal that needs to be sent.

<p class="footer"><a href="#top">top</a>

<p style="text-align:center;">Last modified 9 November 2012</p>

<!--#include virtual="footer.txt"-->