2005-10-02 14:57:44 +00:00
|
|
|
/* -*- C -*-
|
2006-06-08 18:27:17 +00:00
|
|
|
*
|
2005-11-05 19:57:48 +00:00
|
|
|
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
|
|
* University Research and Technology
|
|
|
|
* Corporation. All rights reserved.
|
2006-08-23 03:32:36 +00:00
|
|
|
* Copyright (c) 2004-2006 The University of Tennessee and The University
|
2005-11-05 19:57:48 +00:00
|
|
|
* of Tennessee Research Foundation. All rights
|
|
|
|
* reserved.
|
2006-06-08 18:27:17 +00:00
|
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
2005-10-02 14:57:44 +00:00
|
|
|
* University of Stuttgart. All rights reserved.
|
|
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
|
|
* All rights reserved.
|
|
|
|
* $COPYRIGHT$
|
2006-06-08 18:27:17 +00:00
|
|
|
*
|
2005-10-02 14:57:44 +00:00
|
|
|
* Additional copyrights may follow
|
2006-06-08 18:27:17 +00:00
|
|
|
*
|
2005-10-02 14:57:44 +00:00
|
|
|
* $HEADER$
|
|
|
|
*
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
/**
|
|
|
|
* @file:
|
|
|
|
* Header file for the bproc launcher. This launcher is actually split into 2
|
|
|
|
* modules: pls_bproc & pls_bproc_orted. The general idea behind this launcher is:
|
2006-06-08 18:27:17 +00:00
|
|
|
* -# pls_bproc is called by orterun. It figures out the process mapping and
|
2005-10-02 14:57:44 +00:00
|
|
|
* launches orted's on the nodes
|
|
|
|
* -# pls_bproc_orted is called by orted. This module intializes either a pty or
|
|
|
|
* pipes, places symlinks to them in well know points of the filesystem, and
|
|
|
|
* sets up the io forwarding. It then sends an ack back to orterun.
|
|
|
|
* -# pls_bproc waits for an ack to come back from the orteds, then does several
|
2006-06-08 18:27:17 +00:00
|
|
|
* parallel launches of the application processes. The number of launches is
|
2005-10-02 14:57:44 +00:00
|
|
|
* equal to the maximum number of processes on a node. For example, if there
|
|
|
|
* were 2 processes assigned to node 1, and 1 process asigned to node 2, we
|
2006-06-08 18:27:17 +00:00
|
|
|
* would do a parallel launch that launches on process on each node, then
|
2005-10-02 14:57:44 +00:00
|
|
|
* another which launches another process on node 1.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef ORTE_PLS_BPROC_H_
|
|
|
|
#define ORTE_PLS_BPROC_H_
|
|
|
|
|
|
|
|
#include "orte_config.h"
|
2006-02-12 01:33:29 +00:00
|
|
|
#include "orte/orte_constants.h"
|
2006-09-14 21:29:51 +00:00
|
|
|
|
2005-10-02 14:57:44 +00:00
|
|
|
#include <sys/bproc.h>
|
|
|
|
|
2006-09-14 21:29:51 +00:00
|
|
|
#include "opal/threads/condition.h"
|
|
|
|
|
|
|
|
#include "orte/class/orte_pointer_array.h"
|
|
|
|
#include "orte/util/proc_info.h"
|
|
|
|
|
|
|
|
#include "orte/mca/rml/rml_types.h"
|
|
|
|
|
|
|
|
#include "orte/mca/pls/base/base.h"
|
|
|
|
|
2005-10-02 14:57:44 +00:00
|
|
|
#if defined(c_plusplus) || defined(__cplusplus)
|
|
|
|
extern "C" {
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Module open / close
|
|
|
|
*/
|
|
|
|
int orte_pls_bproc_component_open(void);
|
|
|
|
int orte_pls_bproc_component_close(void);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Startup / Shutdown
|
|
|
|
*/
|
|
|
|
orte_pls_base_module_t* orte_pls_bproc_init(int *priority);
|
|
|
|
int orte_pls_bproc_finalize(void);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Interface
|
|
|
|
*/
|
|
|
|
int orte_pls_bproc_launch(orte_jobid_t);
|
Bring over the update to terminate orteds that are generated by a dynamic spawn such as comm_spawn. This introduces the concept of a job "family" - i.e., jobs that have a parent/child relationship. Comm_spawn'ed jobs have a parent (the one that spawned them). We track that relationship throughout the lineage - i.e., if a comm_spawned job in turn calls comm_spawn, then it has a parent (the one that spawned it) and a "root" job (the original job that started things).
Accordingly, there are new APIs to the name service to support the ability to get a job's parent, root, immediate children, and all its descendants. In addition, the terminate_job, terminate_orted, and signal_job APIs for the PLS have been modified to accept attributes that define the extent of their actions. For example, doing a "terminate_job" with an attribute of ORTE_NS_INCLUDE_DESCENDANTS will terminate the given jobid AND all jobs that descended from it.
I have tested this capability on a MacBook under rsh, Odin under SLURM, and LANL's Flash (bproc). It worked successfully on non-MPI jobs (both simple and including a spawn), and MPI jobs (again, both simple and with a spawn).
This commit was SVN r12597.
2006-11-14 19:34:59 +00:00
|
|
|
int orte_pls_bproc_terminate_job(orte_jobid_t, opal_list_t*);
|
2005-10-02 14:57:44 +00:00
|
|
|
int orte_pls_bproc_terminate_proc(const orte_process_name_t* proc_name);
|
Bring over the update to terminate orteds that are generated by a dynamic spawn such as comm_spawn. This introduces the concept of a job "family" - i.e., jobs that have a parent/child relationship. Comm_spawn'ed jobs have a parent (the one that spawned them). We track that relationship throughout the lineage - i.e., if a comm_spawned job in turn calls comm_spawn, then it has a parent (the one that spawned it) and a "root" job (the original job that started things).
Accordingly, there are new APIs to the name service to support the ability to get a job's parent, root, immediate children, and all its descendants. In addition, the terminate_job, terminate_orted, and signal_job APIs for the PLS have been modified to accept attributes that define the extent of their actions. For example, doing a "terminate_job" with an attribute of ORTE_NS_INCLUDE_DESCENDANTS will terminate the given jobid AND all jobs that descended from it.
I have tested this capability on a MacBook under rsh, Odin under SLURM, and LANL's Flash (bproc). It worked successfully on non-MPI jobs (both simple and including a spawn), and MPI jobs (again, both simple and with a spawn).
This commit was SVN r12597.
2006-11-14 19:34:59 +00:00
|
|
|
int orte_pls_bproc_terminate_orteds(orte_jobid_t jobid, opal_list_t*);
|
|
|
|
int orte_pls_bproc_signal_job(orte_jobid_t, int32_t, opal_list_t*);
|
2006-06-08 18:27:17 +00:00
|
|
|
int orte_pls_bproc_signal_proc(const orte_process_name_t* proc_name, int32_t);
|
2005-10-02 14:57:44 +00:00
|
|
|
|
2006-09-14 21:29:51 +00:00
|
|
|
/* Utility routine to get/set process pid */
|
2006-11-11 04:03:45 +00:00
|
|
|
ORTE_DECLSPEC int orte_pls_bproc_set_proc_pid(const orte_process_name_t*, pid_t, int);
|
2006-09-14 21:29:51 +00:00
|
|
|
ORTE_DECLSPEC int orte_pls_bproc_get_proc_pid(const orte_process_name_t*, pid_t*);
|
|
|
|
/**
|
|
|
|
* Utility routine to retreive all process pids w/in a specified job.
|
|
|
|
*/
|
Bring over the update to terminate orteds that are generated by a dynamic spawn such as comm_spawn. This introduces the concept of a job "family" - i.e., jobs that have a parent/child relationship. Comm_spawn'ed jobs have a parent (the one that spawned them). We track that relationship throughout the lineage - i.e., if a comm_spawned job in turn calls comm_spawn, then it has a parent (the one that spawned it) and a "root" job (the original job that started things).
Accordingly, there are new APIs to the name service to support the ability to get a job's parent, root, immediate children, and all its descendants. In addition, the terminate_job, terminate_orted, and signal_job APIs for the PLS have been modified to accept attributes that define the extent of their actions. For example, doing a "terminate_job" with an attribute of ORTE_NS_INCLUDE_DESCENDANTS will terminate the given jobid AND all jobs that descended from it.
I have tested this capability on a MacBook under rsh, Odin under SLURM, and LANL's Flash (bproc). It worked successfully on non-MPI jobs (both simple and including a spawn), and MPI jobs (again, both simple and with a spawn).
This commit was SVN r12597.
2006-11-14 19:34:59 +00:00
|
|
|
ORTE_DECLSPEC int orte_pls_bproc_get_proc_pids(orte_jobid_t jobid, pid_t** pids,
|
|
|
|
orte_std_cntr_t* num_pids,
|
|
|
|
opal_list_t *attrs);
|
|
|
|
|
2006-09-14 21:29:51 +00:00
|
|
|
/**
|
|
|
|
* Utility routine to get/set daemon pid
|
|
|
|
*/
|
|
|
|
ORTE_DECLSPEC int orte_pls_bproc_set_node_pid(orte_cellid_t cellid, char* node_name, orte_jobid_t jobid, pid_t pid);
|
|
|
|
ORTE_DECLSPEC int orte_pls_bproc_get_node_pids(orte_jobid_t jobid, pid_t** pids, orte_std_cntr_t* num_pids);
|
|
|
|
|
|
|
|
/* utility functions for abort communications */
|
|
|
|
int orte_pls_bproc_comm_start(void);
|
|
|
|
int orte_pls_bproc_comm_stop(void);
|
|
|
|
void orte_pls_bproc_recv(int status, orte_process_name_t* sender,
|
|
|
|
orte_buffer_t* buffer, orte_rml_tag_t tag,
|
|
|
|
void* cbdata);
|
|
|
|
|
2005-10-02 14:57:44 +00:00
|
|
|
/**
|
|
|
|
* PLS bproc Component
|
|
|
|
*/
|
|
|
|
struct orte_pls_bproc_component_t {
|
|
|
|
orte_pls_base_component_t super;
|
|
|
|
/**< The base class */
|
|
|
|
bool done_launching;
|
|
|
|
/**< Is true if we are done launching the user's app. */
|
2006-06-08 18:27:17 +00:00
|
|
|
char * orted;
|
2005-10-02 14:57:44 +00:00
|
|
|
/**< The orted executeable. This can be an absolute path, or if not found
|
|
|
|
* we will look for it in the user's path */
|
|
|
|
int debug;
|
|
|
|
/**< If greater than 0 print debugging information */
|
Bring the timing instrumentation to the trunk.
If you want to look at our launch and MPI process startup times, you can do so with two MCA params:
OMPI_MCA_orte_timing: set it to anything non-zero and you will get the launch time for different steps in the job launch procedure. The degree of detail depends on the launch environment. rsh will provide you with the average, min, and max launch time for the daemons. SLURM block launches the daemon, so you only get the time to launch the daemons and the total time to launch the job. Ditto for bproc. TM looks more like rsh. Only those four environments are currently supported - anyone interested in extending this capability to other environs is welcome to do so. In all cases, you also get the time to setup the job for launch.
OMPI_MCA_ompi_timing: set it to anything non-zero and you will get the time for mpi_init to reach the compound registry command, the time to execute that command, the time to go from our stage1 barrier to the stage2 barrier, and the time to go from the stage2 barrier to the end of mpi_init. This will be output for each process, so you'll have to compile any statistics on your own. Note: if someone develops a nice parser to do so, it would be really appreciated if you could/would share!
This commit was SVN r12302.
2006-10-25 15:27:47 +00:00
|
|
|
bool timing;
|
|
|
|
/**< If true, report launch timing info */
|
2005-10-02 14:57:44 +00:00
|
|
|
int num_procs;
|
|
|
|
/**< The number of processes that are running */
|
|
|
|
int priority;
|
|
|
|
/**< The priority of this component. This will be returned if we determine
|
|
|
|
* that bproc is available and running on this node, */
|
|
|
|
int terminate_sig;
|
|
|
|
/**< The signal that gets sent to a process to kill it. */
|
|
|
|
size_t num_daemons;
|
|
|
|
/**< The number of daemons that are currently running. */
|
|
|
|
opal_mutex_t lock;
|
|
|
|
/**< Lock used to prevent some race conditions */
|
|
|
|
opal_condition_t condition;
|
|
|
|
/**< Condition that is signaled when all the daemons have died */
|
|
|
|
orte_pointer_array_t * daemon_names;
|
|
|
|
/**< Array of the process names of all the daemons. This is used to send
|
|
|
|
* the daemons a termonation signal when all the user processes are done */
|
2006-08-03 05:29:49 +00:00
|
|
|
orte_pointer_array_t* active_node_names;
|
|
|
|
/**< Array of the bproc node names of all the daemons. This is used to
|
|
|
|
* track which bproc nodes belong to us*/
|
2006-08-11 19:41:33 +00:00
|
|
|
bool bynode;
|
|
|
|
/**< Indicates whether or not this application is to be mapped by node
|
|
|
|
* (if set to true) or by slot (default)
|
|
|
|
*/
|
2006-09-14 21:29:51 +00:00
|
|
|
bool recv_issued;
|
|
|
|
/**< Indicates that the comm recv for reporting abnormal proc termination
|
|
|
|
* has been issued
|
|
|
|
*/
|
2006-08-03 05:29:49 +00:00
|
|
|
|
2005-10-02 14:57:44 +00:00
|
|
|
};
|
|
|
|
/**
|
|
|
|
* Convenience typedef
|
|
|
|
*/
|
|
|
|
typedef struct orte_pls_bproc_component_t orte_pls_bproc_component_t;
|
|
|
|
|
2006-08-23 03:32:36 +00:00
|
|
|
ORTE_DECLSPEC orte_pls_bproc_component_t mca_pls_bproc_component;
|
|
|
|
ORTE_DECLSPEC orte_pls_base_module_t orte_pls_bproc_module;
|
2005-10-02 14:57:44 +00:00
|
|
|
|
|
|
|
#if defined(c_plusplus) || defined(__cplusplus)
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
#endif /* ORTE_PLS_BPROC_H_ */
|
|
|
|
|