1
1
openmpi/orte/mca/plm/ccp/plm_ccp_module.c

740 строки
24 KiB
C

/*
* Copyright (c) 2004-2007 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2004-2008 High Performance Computing Center Stuttgart,
* University of Stuttgart. All rights reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
*
* $HEADER$
*
*/
#include "orte_config.h"
#include "orte/constants.h"
#include "orte/types.h"
#if HAVE_UNISTD_H
#include <unistd.h>
#endif
#include <signal.h>
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#ifdef HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
#ifdef HAVE_SYS_WAIT_H
#include <sys/wait.h>
#endif
#ifdef HAVE_SCHED_H
#include <sched.h>
#endif
#ifdef HAVE_SYS_TIME_H
#include <sys/time.h>
#endif
#include <errno.h>
#include <comutil.h>
#include "opal/mca/installdirs/installdirs.h"
#include "opal/threads/condition.h"
#include "opal/event/event.h"
#include "opal/util/argv.h"
#include "opal/util/output.h"
#include "opal/util/opal_environ.h"
#include "opal/util/show_help.h"
#include "opal/util/path.h"
#include "opal/util/basename.h"
#include "opal/mca/base/mca_base_param.h"
#include "opal/runtime/opal_progress.h"
#include "orte/util/name_fns.h"
#include "orte/runtime/orte_globals.h"
#include "orte/runtime/orte_wait.h"
#include "orte/mca/errmgr/errmgr.h"
#include "orte/mca/rmaps/rmaps.h"
#include "orte/mca/plm/plm.h"
#include "orte/mca/plm/base/plm_private.h"
#include "plm_ccp.h"
/* Import the Windows CCP API. */
#import "ccpapi.tlb" named_guids no_namespace raw_interfaces_only \
rename("SetEnvironmentVariable","SetEnvVar") \
rename("GetJob", "GetSingleJob") \
rename("AddJob", "AddSingleJob")
/*
* Local functions
*/
static int plm_ccp_init(void);
static int plm_ccp_launch_job(orte_job_t *jdata);
static int plm_ccp_terminate_job(orte_jobid_t jobid);
static int plm_ccp_terminate_orteds();
static int plm_ccp_signal_job(orte_jobid_t jobid, int32_t signal);
static int plm_ccp_finalize(void);
static int plm_ccp_connect(ICluster* pCluster);
static int plm_ccp_disconnect(void);
void get_cluster_message(ICluster* pCluster);
static char *plm_ccp_commandline(char *node_name, int argc, char **argv);
/*
* Global variable
*/
orte_plm_base_module_t orte_plm_ccp_module = {
plm_ccp_init,
orte_plm_base_set_hnp_name,
plm_ccp_launch_job,
plm_ccp_terminate_job,
plm_ccp_terminate_orteds,
plm_ccp_signal_job,
plm_ccp_finalize
};
/**
* Init the module
*/
static int plm_ccp_init(void)
{
int rc;
if (ORTE_SUCCESS != (rc = orte_plm_base_comm_start())) {
ORTE_ERROR_LOG(rc);
}
return rc;
}
/* When working in this function, ALWAYS jump to "cleanup" if
* you encounter an error so that orterun will be woken up and
* the job can cleanly terminate
*/
static int plm_ccp_launch_job(orte_job_t *jdata)
{
orte_app_context_t **apps;
orte_node_t **nodes;
orte_std_cntr_t launched = 0, i;
int local_err;
orte_job_map_t *map = NULL;
opal_list_item_t *item;
size_t num_nodes;
int argc, rc, node_name_index, proc_vpid_index, proc_name_index;
char *param, **env = NULL, *var, **argv = NULL;
bool connected = false;
char *bin_base = NULL, *lib_base = NULL, command_line[300];
struct timeval completionstart, completionstop, launchstart, launchstop;
struct timeval jobstart, jobstop;
int maxtime=0, mintime=99999999, maxiter = 0, miniter = 0, deltat;
float avgtime=0.0;
bool failed_launch = true;
mode_t current_umask;
IClusterEnumerable* pNodesCollection = NULL;
IEnumVARIANT* pNodes = NULL;
VARIANT v;
INode* pNode = NULL;
HRESULT hr = S_OK;
ICluster* pCluster = NULL;
IJob* pJob = NULL;
long job_id, num_processors = 0, idle_processors = 0;
IClusterCounter* pClusterCounter = NULL;
ITask* pTask = NULL;
JobPriority job_priority = JobPriority_Normal;
/* check for timing request - get start time if so */
if (orte_timing) {
if (0 != gettimeofday(&jobstart, NULL)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: could not obtain job start time"));
}
}
/* if this jobid isn't invalid, then it already
* has been setup, so skip the setup actions
*/
if (ORTE_JOBID_INVALID != jdata->jobid) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: launching job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jdata->jobid)));
goto GETMAP;
}
/* create a jobid for this job */
if (ORTE_SUCCESS != (rc = orte_plm_base_create_jobid(&jdata->jobid))) {
ORTE_ERROR_LOG(rc);
goto cleanup;
}
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: launching job %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jdata->jobid)));
/* setup the job */
if (ORTE_SUCCESS != (rc = orte_plm_base_setup_job(jdata))) {
ORTE_ERROR_LOG(rc);
goto cleanup;
}
GETMAP:
/* Get the map for this job */
if (NULL == (map = orte_rmaps.get_job_map(jdata->jobid))) {
ORTE_ERROR_LOG(ORTE_ERR_NOT_FOUND);
rc = ORTE_ERR_NOT_FOUND;
goto cleanup;
}
apps = (orte_app_context_t**)jdata->apps->addr;
nodes = (orte_node_t**)map->nodes->addr;
if (0 == map->num_new_daemons) {
/* have all the daemons we need - launch app */
goto launch_apps;
}
/* add the daemon command (as specified by user) */
argv = opal_argv_split(mca_plm_ccp_component.orted, ' ');
argc = opal_argv_count(argv);
opal_argv_append(&argc, &argv, "--no-daemonize");
/* Add basic orted command line options */
orte_plm_base_orted_append_basic_args(&argc, &argv, "env",
&proc_vpid_index,
&node_name_index);
if (0 < opal_output_get_verbosity(orte_plm_globals.output)) {
param = opal_argv_join(argv, ' ');
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: final top-level argv:\n\t%s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
(NULL == param) ? "NULL" : param));
if (NULL != param) free(param);
}
/* CCP is not thread safe. Use the apartment model. */
CoInitializeEx(NULL, COINIT_APARTMENTTHREADED);
/* Create the Cluster object. */
hr = CoCreateInstance( __uuidof(Cluster),
NULL,
CLSCTX_INPROC_SERVER,
__uuidof(ICluster),
reinterpret_cast<void **> (&pCluster) );
if (FAILED(hr)) {
opal_output(orte_plm_globals.output,
"plm:ccp: failed to create cluster object!");
goto cleanup;
}
/* Connect to the head node. */
rc = plm_ccp_connect(pCluster);
if (ORTE_SUCCESS != rc) {
goto cleanup;
}
connected = true;
hr = pCluster->CreateJob(&pJob);
if (FAILED(hr)) {
get_cluster_message(pCluster);
opal_output(orte_plm_globals.output,
"plm:ccp:failed to create cluster object!");
goto cleanup;
}
/* Figure out the basenames for the libdir and bindir. There is a
lengthy comment about this in plm_rsh_module.c explaining all
the rationale for how / why we're doing this. */
lib_base = opal_basename(opal_install_dirs.libdir);
bin_base = opal_basename(opal_install_dirs.bindir);
/* setup environment */
env = opal_argv_copy(environ);
/* add our umask -- see big note in orted.c */
current_umask = umask(0);
umask(current_umask);
asprintf(&var, "0%o", current_umask);
opal_setenv("ORTE_DAEMON_UMASK_VALUE", var, true, &env);
free(var);
/* If we have a prefix, then modify the PATH and
LD_LIBRARY_PATH environment variables. We only allow
a single prefix to be specified. Since there will
always be at least one app_context, we take it from
there
*/
if (NULL != apps[0]->prefix_dir) {
char *newenv;
for (i = 0; NULL != env && NULL != env[i]; ++i) {
/* Reset PATH */
if (0 == strncmp("PATH=", env[i], 5)) {
asprintf(&newenv, "%s/%s:%s",
apps[0]->prefix_dir, bin_base, env[i] + 5);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: resetting PATH: %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
newenv));
opal_setenv("PATH", newenv, true, &env);
free(newenv);
}
/* Reset LD_LIBRARY_PATH */
else if (0 == strncmp("LD_LIBRARY_PATH=", env[i], 16)) {
asprintf(&newenv, "%s/%s:%s",
apps[0]->prefix_dir, lib_base, env[i] + 16);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: resetting LD_LIBRARY_PATH: %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
newenv));
opal_setenv("LD_LIBRARY_PATH", newenv, true, &env);
free(newenv);
}
}
}
/* This has already been done in RAS, but I have to do it again here.
* Because the node structure doesn't have num_processor member. */
/* Get the collection of nodes. */
hr = pCluster->get_ComputeNodes(&pNodesCollection);
/* Get the enumerator used to iterate through the collection. */
hr = pNodesCollection->GetEnumerator(&pNodes);
VariantInit(&v);
/* Loop through the collection. */
while (hr = pNodes->Next(1, &v, NULL) == S_OK) {
v.pdispVal->QueryInterface(IID_INode, reinterpret_cast<void **> (&pNode));
/* Iterate through each of the nodes and check to sum up all the processors. */
for (i = 0; i < map->num_nodes; i++) {
orte_node_t* node = nodes[i];
char* vpid_string;
BSTR node_name;
hr = pNode->get_Name(&node_name);
if( 0 == strcmp(_com_util::ConvertBSTRToString(node_name), node->name)) {
/* Get available number of processors on required node. */
hr = pNode->get_NumberOfIdleProcessors(&idle_processors);
num_processors += idle_processors;
}
}
}
pJob->put_MinimumNumberOfProcessors(num_processors);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to put min num of processors!"));
goto cleanup;
}
pJob->put_MaximumNumberOfProcessors(num_processors);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to put max num of processors!"));
goto cleanup;
}
hr = pJob->put_Priority(job_priority);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to set proiority!"));
goto cleanup;
}
hr = pJob->SetExtendedJobTerm(_bstr_t(L"extended terms"), _bstr_t(L"TermValue"));
/* Iterate through each of the nodes and spin
* up a daemon.
*/
for (i = 0; i < map->num_nodes; i++) {
orte_node_t* node = nodes[i];
char* vpid_string;
/* if this daemon already exists, don't launch it! */
if (node->daemon_launched) {
continue;
}
/* setup node name */
free(argv[node_name_index]);
argv[node_name_index] = strdup(node->name);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: launching on node %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
node->name));
/* setup process name */
rc = orte_util_convert_vpid_to_string(&vpid_string, nodes[i]->daemon->name.vpid);
if (ORTE_SUCCESS != rc) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp: unable to get daemon vpid as string"));
exit(-1);
}
free(argv[proc_vpid_index]);
argv[proc_vpid_index] = strdup(vpid_string);
free(vpid_string);
/* exec the daemon */
if (0 < opal_output_get_verbosity(orte_plm_globals.output)) {
param = opal_argv_join(argv, ' ');
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: executing:\n\t%s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
(NULL == param) ? "NULL" : param));
if (NULL != param) free(param);
}
/* check for timing request - get start time if so */
if (orte_timing) {
if (0 != gettimeofday(&launchstart, NULL)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: could not obtain start time"));
launchstart.tv_sec = 0;
launchstart.tv_usec = 0;
}
}
/* Set terms for task. */
hr = pCluster->CreateTask(&pTask);
if (FAILED(hr)) {
get_cluster_message(pCluster);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to create task object!"));
goto cleanup;
}
pTask->put_MinimumNumberOfProcessors(node->num_procs);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to create task object!"));
goto cleanup;
}
pTask->put_MaximumNumberOfProcessors(node->num_procs);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to create task object!"));
goto cleanup;
}
pTask->put_RequiredNodes(_bstr_t(node->name));
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to set required nodes!"));
goto cleanup;
}
/* Prepare the command a little bit. */
strcpy_s(command_line, apps[0]->prefix_dir);
strcat(command_line,"\\");
{
char* ccp_cmdline = plm_ccp_commandline(node->name, argc, argv);
strcat(command_line, ccp_cmdline);
free(ccp_cmdline);
}
hr = pTask->put_CommandLine(_bstr_t(command_line));
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to put command line!"));
goto cleanup;
}
hr = pTask->put_Stdout(_bstr_t(L"ompi_ccp_output.txt"));
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to set stdout!"));
goto cleanup;
}
hr = pTask->put_Stderr(_bstr_t(L"ompi_ccp_error.txt"));
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to set stderr!"));
goto cleanup;
}
hr = pJob->AddTask(pTask);
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:failed to add task!"));
goto cleanup;
}
/* Allow some progress to occur */
opal_event_loop(OPAL_EVLOOP_NONBLOCK);
launched++;
pTask->Release();
}
/* Add job to the queue. */
hr = pCluster->QueueJob(pJob, NULL, NULL, VARIANT_TRUE, 0, &job_id);
if (SUCCEEDED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"Added job %d to scheduling queue.\n", job_id));
}else {
get_cluster_message(pCluster);
}
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp:launch: finished spawning orteds",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
/* wait for daemons to callback */
if (ORTE_SUCCESS != (rc = orte_plm_base_daemon_callback(map->num_new_daemons))) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: daemon launch failed for job %s on error %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jdata->jobid), ORTE_ERROR_NAME(rc)));
goto cleanup;
}
launch_apps:
if (ORTE_SUCCESS != (rc = orte_plm_base_launch_apps(jdata->jobid))) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp: launch of apps failed for job %s on error %s",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME),
ORTE_JOBID_PRINT(jdata->jobid), ORTE_ERROR_NAME(rc)));
goto cleanup;
}
/* if we get here, then everything launched okay - record that fact */
failed_launch = false;
/* check for timing request - get stop time for launch completion and report */
if (orte_timing) {
if (0 != gettimeofday(&completionstop, NULL)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: could not obtain completion stop time"));
} else {
deltat = (launchstop.tv_sec - launchstart.tv_sec)*1000000 +
(launchstop.tv_usec - launchstart.tv_usec);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: launch completion required %d usec", deltat));
}
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: Launch statistics:"));
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: Average time to launch an orted: %f usec", avgtime));
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: Max time to launch an orted: %d usec at iter %d", maxtime, maxiter));
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: Min time to launch an orted: %d usec at iter %d", mintime, miniter));
}
cleanup:
if (NULL != argv) {
opal_argv_free(argv);
}
if (NULL != env) {
opal_argv_free(env);
}
if (connected) {
plm_ccp_disconnect();
}
if (NULL != lib_base) {
free(lib_base);
}
if (NULL != bin_base) {
free(bin_base);
}
/* check for failed launch - if so, force terminate */
if (failed_launch) {
orte_plm_base_launch_failed(jdata->jobid, false, -1, ORTE_ERROR_DEFAULT_EXIT_CODE, ORTE_JOB_STATE_FAILED_TO_START);
}
/* check for timing request - get stop time and process if so */
if (orte_timing) {
if (0 != gettimeofday(&jobstop, NULL)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: could not obtain stop time"));
} else {
deltat = (jobstop.tv_sec - jobstart.tv_sec)*1000000 +
(jobstop.tv_usec - jobstart.tv_usec);
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm_ccp: launch of entire job required %d usec", deltat));
}
}
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"%s plm:ccp:launch: finished",
ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
return rc;
}
static int plm_ccp_terminate_job(orte_jobid_t jobid)
{
int rc;
/* order all of the daemons to kill their local procs for this job */
if (ORTE_SUCCESS != (rc = orte_plm_base_orted_kill_local_procs(jobid))) {
ORTE_ERROR_LOG(rc);
}
return rc;
}
/**
* Terminate the orteds for a given job
*/
int plm_ccp_terminate_orteds()
{
int rc;
/* now tell them to die! */
if (ORTE_SUCCESS != (rc = orte_plm_base_orted_exit())) {
ORTE_ERROR_LOG(rc);
}
return rc;
}
static int plm_ccp_signal_job(orte_jobid_t jobid, int32_t signal)
{
int rc;
/* order them to pass this signal to their local procs */
if (ORTE_SUCCESS != (rc = orte_plm_base_orted_signal_local_procs(jobid, signal))) {
ORTE_ERROR_LOG(rc);
}
return rc;
}
/*
* Free stuff
*/
static int plm_ccp_finalize(void)
{
int rc;
/* cleanup any pending recvs */
if (ORTE_SUCCESS != (rc = orte_plm_base_comm_stop())) {
ORTE_ERROR_LOG(rc);
}
return ORTE_SUCCESS;
}
static int plm_ccp_connect(ICluster* pCluster)
{
size_t i, len;
char *cluster_head = NULL;
HRESULT hr = S_OK;
/* Get the cluster head nodes name */
_dupenv_s(&cluster_head, &len, "LOGONSERVER");
if(cluster_head == NULL) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:allocate: connot find cluster head node!"));
return ORTE_ERROR;
}
/* Get rid of the beginning '//'. */
for( i = 0; i < len; i++){
cluster_head[i] = cluster_head[i+2];
cluster_head[i+2] = '\0';
}
/* Connect to the cluster's head node */
hr = pCluster->Connect(_bstr_t(cluster_head));
if (FAILED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"plm:ccp:allocate: connection failed!"));
return ORTE_ERROR;
}
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"Connected to Cluster: %s. \n", cluster_head));
return ORTE_SUCCESS;
}
static int plm_ccp_disconnect(void)
{
return ORTE_SUCCESS;
}
/* Generate the proper command line according to the env. */
static char *plm_ccp_commandline(char *node_name, int argc, char **argv)
{
char *commandline;
int i, len = 0;
for( i = 0; i < argc; i++ ) {
len += strlen(argv[i]) + 1;
}
commandline = (char*)malloc( len + 1);
memset(commandline, '\0', len+1);
for(i=0;i<argc;i++) {
/* Don't know why we use these -mca args, I have to ignore them
* otherwise the command line will be too long for CCP. */
if( 0 == strcmp("-mca", argv[i]) &&
(0 == strcmp("mca_base_param_file_path", argv[i+1]) ||
0 == strcmp("mca_base_param_file_path_force", argv[i+1])) ) {
i += 2;
continue;
}
/* Unknown option "--no-daemonize" for Windows? */
if ( 0 == strcmp("--no-daemonize", argv[i]) ) {
continue;
}
/* Append command args, and separate them with spaces. */
strcat(commandline, argv[i]);
commandline[strlen(commandline)]=' ';
}
return commandline;
}
void get_cluster_message(ICluster* pCluster)
{
HRESULT hr = S_OK;
BSTR message = NULL;
hr = pCluster->get_ErrorMessage(&message);
if (SUCCEEDED(hr)) {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
_com_util::ConvertBSTRToString(message)));
SysFreeString(message);
}
else {
OPAL_OUTPUT_VERBOSE((1, orte_plm_globals.output,
"pCluster->get_ErrorMessage failed.\n"));
}
}