2004-10-15 20:31:54 +04:00
|
|
|
/*
|
2005-11-05 22:57:48 +03:00
|
|
|
* Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
|
|
|
|
* University Research and Technology
|
|
|
|
* Corporation. All rights reserved.
|
2006-08-23 07:32:36 +04:00
|
|
|
* Copyright (c) 2004-2006 The University of Tennessee and The University
|
2005-11-05 22:57:48 +03:00
|
|
|
* of Tennessee Research Foundation. All rights
|
|
|
|
* reserved.
|
2004-11-28 23:09:25 +03:00
|
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
|
|
* University of Stuttgart. All rights reserved.
|
2005-03-24 15:43:37 +03:00
|
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
|
|
* All rights reserved.
|
2004-11-22 04:38:40 +03:00
|
|
|
* $COPYRIGHT$
|
|
|
|
*
|
|
|
|
* Additional copyrights may follow
|
|
|
|
*
|
2004-10-15 20:31:54 +04:00
|
|
|
* $HEADER$
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
2005-03-14 23:57:21 +03:00
|
|
|
#include "orte_config.h"
|
2008-02-28 04:57:57 +03:00
|
|
|
#include "orte/constants.h"
|
2004-10-15 20:31:54 +04:00
|
|
|
|
|
|
|
#include <stdio.h>
|
2004-10-20 05:03:09 +04:00
|
|
|
#ifdef HAVE_PWD_H
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <pwd.h>
|
2004-10-20 05:03:09 +04:00
|
|
|
#endif
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <stddef.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
2004-10-20 05:03:09 +04:00
|
|
|
#ifdef HAVE_SYS_PARAM_H
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <sys/param.h>
|
2007-04-01 20:16:54 +04:00
|
|
|
#endif /* HAVE_SYS_PARAM_H */
|
2004-10-20 05:03:09 +04:00
|
|
|
#ifdef HAVE_SYS_TYPES_H
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <sys/types.h>
|
2007-04-01 20:16:54 +04:00
|
|
|
#endif /* HAVE_SYS_TYPES_H */
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <sys/stat.h>
|
2004-10-20 05:03:09 +04:00
|
|
|
#ifdef HAVE_UNISTD_H
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <unistd.h>
|
2007-04-01 20:16:54 +04:00
|
|
|
#endif /* HAVE_UNISTD_H */
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <errno.h>
|
2004-10-20 05:03:09 +04:00
|
|
|
#ifdef HAVE_DIRENT_H
|
2004-10-15 20:31:54 +04:00
|
|
|
#include <dirent.h>
|
2007-04-01 20:16:54 +04:00
|
|
|
#endif /* HAVE_DIRENT_H */
|
2008-03-24 02:10:15 +03:00
|
|
|
#ifdef HAVE_PWD_H
|
|
|
|
#include <pwd.h>
|
|
|
|
#endif /* HAVE_PWD_H */
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2008-06-09 18:53:58 +04:00
|
|
|
#include "orte/util/show_help.h"
|
2005-07-04 05:59:52 +04:00
|
|
|
#include "opal/util/os_path.h"
|
2006-07-04 02:23:07 +04:00
|
|
|
#include "opal/util/os_dirpath.h"
|
2006-08-23 07:32:36 +04:00
|
|
|
#include "opal/util/basename.h"
|
2008-03-31 21:10:08 +04:00
|
|
|
#include "opal/util/opal_environ.h"
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2008-02-28 04:57:57 +03:00
|
|
|
#include "orte/util/proc_info.h"
|
|
|
|
#include "orte/util/name_fns.h"
|
|
|
|
|
2006-02-12 04:33:29 +03:00
|
|
|
#include "orte/mca/errmgr/errmgr.h"
|
|
|
|
#include "orte/runtime/runtime.h"
|
2008-02-28 04:57:57 +03:00
|
|
|
#include "orte/runtime/orte_globals.h"
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-02-12 04:33:29 +03:00
|
|
|
#include "orte/util/session_dir.h"
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*******************************
|
|
|
|
* Local function Declarations
|
|
|
|
*******************************/
|
|
|
|
static int orte_create_dir(char *directory);
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
static bool orte_dir_check_file(const char *root, const char *path);
|
2008-02-28 04:57:57 +03:00
|
|
|
static bool orte_dir_check_file_output(const char *root, const char *path);
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2005-06-27 03:11:37 +04:00
|
|
|
#define OMPI_PRINTF_FIX_STRING(a) ((NULL == a) ? "(null)" : a)
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/****************************
|
|
|
|
* Funcationality
|
|
|
|
****************************/
|
|
|
|
/*
|
|
|
|
* Check and create the directory requested
|
|
|
|
*/
|
|
|
|
static int orte_create_dir(char *directory)
|
|
|
|
{
|
2007-04-01 20:16:54 +04:00
|
|
|
mode_t my_mode = S_IRWXU; /* I'm looking for full rights */
|
2006-06-29 01:03:31 +04:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
/* Sanity check before creating the directory with the proper mode,
|
|
|
|
* Make sure it doesn't exist already */
|
2007-07-18 23:50:54 +04:00
|
|
|
if( ORTE_ERR_NOT_FOUND != (ret = opal_os_dirpath_access(directory, my_mode)) ) {
|
2006-07-04 02:23:07 +04:00
|
|
|
/* Failure because opal_os_dirpath_access() indicated that either:
|
2006-06-29 01:03:31 +04:00
|
|
|
* - The directory exists and we can access it (no need to create it again),
|
2006-07-04 02:23:07 +04:00
|
|
|
* return OPAL_SUCCESS, or
|
|
|
|
* - don't have access rights, return OPAL_ERROR
|
2006-06-29 01:03:31 +04:00
|
|
|
*/
|
2007-07-18 23:50:54 +04:00
|
|
|
if (ORTE_SUCCESS != ret) {
|
|
|
|
ORTE_ERROR_LOG(ret);
|
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
return(ret);
|
|
|
|
}
|
2007-07-18 23:50:54 +04:00
|
|
|
|
|
|
|
/* Get here if the directory doesn't exist, so create it */
|
|
|
|
if (ORTE_SUCCESS != (ret = opal_os_dirpath_create(directory, my_mode))) {
|
|
|
|
ORTE_ERROR_LOG(ret);
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2007-07-18 23:50:54 +04:00
|
|
|
return ret;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Construct the fullpath to the session directory
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
orte_session_dir_get_name(char **fulldirpath,
|
2007-04-23 22:25:36 +04:00
|
|
|
char **return_prefix, /* This will come back as the valid tmp dir */
|
|
|
|
char **return_frontend,
|
2008-03-24 02:10:15 +03:00
|
|
|
char *hostid,
|
2008-02-28 04:57:57 +03:00
|
|
|
char *batchid,
|
2006-06-29 01:03:31 +04:00
|
|
|
char *job, char *proc) {
|
|
|
|
char *hostname = NULL,
|
|
|
|
*batchname = NULL,
|
|
|
|
*sessions = NULL,
|
|
|
|
*user = NULL,
|
2007-04-23 22:25:36 +04:00
|
|
|
*prefix = NULL,
|
|
|
|
*frontend = NULL;
|
|
|
|
bool prefix_provided = false;
|
2006-06-29 01:03:31 +04:00
|
|
|
int exit_status = ORTE_SUCCESS;
|
2008-03-24 02:10:15 +03:00
|
|
|
#ifndef __WINDOWS__
|
|
|
|
int uid;
|
|
|
|
struct passwd *pwdent;
|
|
|
|
#else
|
|
|
|
#define INFO_BUF_SIZE 256
|
|
|
|
TCHAR info_buf[INFO_BUF_SIZE];
|
|
|
|
DWORD info_buf_length = INFO_BUF_SIZE;
|
|
|
|
#endif
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/* Ensure that system info is set */
|
2008-03-24 02:10:15 +03:00
|
|
|
orte_proc_info();
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2008-03-24 02:10:15 +03:00
|
|
|
/* get the name of the user */
|
|
|
|
#ifndef __WINDOWS__
|
|
|
|
uid = getuid();
|
|
|
|
#ifdef HAVE_GETPWUID
|
|
|
|
pwdent = getpwuid(uid);
|
|
|
|
#else
|
|
|
|
pwdent = NULL;
|
|
|
|
#endif
|
|
|
|
if (NULL != pwdent) {
|
|
|
|
user = strdup(pwdent->pw_name);
|
|
|
|
} else {
|
|
|
|
if (0 > asprintf(&user, "%d", uid)) {
|
|
|
|
return ORTE_ERR_OUT_OF_RESOURCE;
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2008-03-24 02:10:15 +03:00
|
|
|
#else
|
|
|
|
if (!GetUserName(info_buf, &info_buf_length)) {
|
|
|
|
user = strdup("unknown");
|
|
|
|
} else {
|
|
|
|
user = strdup(info_buf);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* set the 'hostname'
|
|
|
|
*/
|
|
|
|
if( NULL != hostid) { /* User specified version */
|
|
|
|
hostname = strdup(hostid);
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
else { /* check if it is set elsewhere */
|
2008-03-24 02:10:15 +03:00
|
|
|
if( NULL != orte_process_info.nodename)
|
|
|
|
hostname = strdup(orte_process_info.nodename);
|
2006-06-29 01:03:31 +04:00
|
|
|
else {
|
|
|
|
/* Couldn't find it, so fail */
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_BAD_PARAM);
|
|
|
|
exit_status = ORTE_ERR_BAD_PARAM;
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* set the 'batchid'
|
|
|
|
*/
|
|
|
|
if (NULL != batchid)
|
|
|
|
batchname = strdup(batchid);
|
|
|
|
else
|
|
|
|
batchname = strdup("0");
|
|
|
|
|
|
|
|
/*
|
2008-02-28 04:57:57 +03:00
|
|
|
* Check: Can't give a proc without a job
|
2006-06-29 01:03:31 +04:00
|
|
|
*/
|
|
|
|
if( NULL == job &&
|
|
|
|
NULL != proc) {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_BAD_PARAM);
|
|
|
|
exit_status = ORTE_ERR_BAD_PARAM;
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* get the front part of the session directory
|
|
|
|
* Will look something like:
|
|
|
|
* openmpi-sessions-USERNAME@HOSTNAME_BATCHID
|
|
|
|
*/
|
|
|
|
if (NULL != orte_process_info.top_session_dir) {
|
2007-04-23 22:25:36 +04:00
|
|
|
frontend = strdup(orte_process_info.top_session_dir);
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
|
|
|
else { /* If not set then construct it */
|
2007-04-23 22:25:36 +04:00
|
|
|
if (0 > asprintf(&frontend, "openmpi-sessions-%s@%s_%s", user, hostname, batchname)) {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
|
|
|
|
exit_status = ORTE_ERR_OUT_OF_RESOURCE;
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Construct the session directory
|
|
|
|
*/
|
|
|
|
/* If we were given a 'proc' then we can construct it fully into:
|
2008-02-28 04:57:57 +03:00
|
|
|
* openmpi-sessions-USERNAME@HOSTNAME_BATCHID/JOBID/PROC
|
2006-06-29 01:03:31 +04:00
|
|
|
*/
|
|
|
|
if( NULL != proc) {
|
2008-02-28 04:57:57 +03:00
|
|
|
sessions = opal_os_path( false, frontend, job, proc, NULL );
|
2006-08-23 07:32:36 +04:00
|
|
|
if( NULL == sessions ) {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERROR);
|
2006-06-29 01:03:31 +04:00
|
|
|
exit_status = ORTE_ERROR;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/* If we were given a 'job' then we can construct it partially into:
|
2008-02-28 04:57:57 +03:00
|
|
|
* openmpi-sessions-USERNAME@HOSTNAME_BATCHID/JOBID
|
2006-06-29 01:03:31 +04:00
|
|
|
*/
|
|
|
|
else if(NULL != job) {
|
2008-02-28 04:57:57 +03:00
|
|
|
sessions = opal_os_path( false, frontend, job, NULL );
|
2006-08-23 07:32:36 +04:00
|
|
|
if( NULL == sessions ) {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERROR);
|
2006-06-29 01:03:31 +04:00
|
|
|
exit_status = ORTE_ERROR;
|
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
/* If we were not given either then we just set it to frontend
|
2006-07-11 20:54:07 +04:00
|
|
|
*/
|
|
|
|
else {
|
2008-02-28 04:57:57 +03:00
|
|
|
sessions = strdup(frontend); /* must dup this to avoid double-free later */
|
2006-07-11 20:54:07 +04:00
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If the user specified an invalid prefix, or no prefix at all
|
|
|
|
* we need to keep looking
|
|
|
|
*/
|
2008-02-28 04:57:57 +03:00
|
|
|
if( NULL != fulldirpath && NULL != *fulldirpath) {
|
2006-06-29 01:03:31 +04:00
|
|
|
free(*fulldirpath);
|
|
|
|
*fulldirpath = NULL;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2007-04-23 22:25:36 +04:00
|
|
|
if( NULL != return_prefix && NULL != *return_prefix) { /* use the user specified one, if available */
|
|
|
|
prefix = strdup(*return_prefix);
|
|
|
|
prefix_provided = true;
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
|
|
|
/* Try to find a proper alternative prefix */
|
|
|
|
else if (NULL != orte_process_info.tmpdir_base) { /* stored value */
|
2007-04-23 22:25:36 +04:00
|
|
|
prefix = strdup(orte_process_info.tmpdir_base);
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
|
|
|
else if( NULL != getenv("OMPI_PREFIX_ENV") ) { /* OMPI Environment var */
|
2007-04-23 22:25:36 +04:00
|
|
|
prefix = strdup(getenv("OMPI_PREFIX_ENV"));
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
2008-03-31 21:10:08 +04:00
|
|
|
else { /* General Environment var */
|
|
|
|
prefix = strdup(opal_tmp_directory());
|
2006-06-29 01:03:31 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2008-02-28 04:57:57 +03:00
|
|
|
* Construct the absolute final path, if requested
|
2006-06-29 01:03:31 +04:00
|
|
|
*/
|
2008-02-28 04:57:57 +03:00
|
|
|
if (NULL != fulldirpath) {
|
|
|
|
*fulldirpath = opal_os_path(false, prefix, sessions, NULL);
|
|
|
|
}
|
2007-04-23 22:25:36 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Return the frontend and prefix, if user requested we do so
|
|
|
|
*/
|
|
|
|
if (NULL != return_frontend) {
|
|
|
|
*return_frontend = strdup(frontend);
|
|
|
|
}
|
|
|
|
if (!prefix_provided && NULL != return_prefix) {
|
|
|
|
*return_prefix = strdup(prefix);
|
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
cleanup:
|
|
|
|
if(NULL != hostname)
|
|
|
|
free(hostname);
|
|
|
|
if(NULL != batchname)
|
|
|
|
free(batchname);
|
|
|
|
if(NULL != sessions)
|
|
|
|
free(sessions);
|
|
|
|
if(NULL != user)
|
|
|
|
free(user);
|
2007-04-23 22:25:36 +04:00
|
|
|
if (NULL != prefix) free(prefix);
|
|
|
|
if (NULL != frontend) free(frontend);
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
return exit_status;
|
|
|
|
}
|
2005-03-19 02:58:36 +03:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Construct the session directory and create it if necessary
|
|
|
|
*/
|
|
|
|
int orte_session_dir(bool create,
|
2008-03-24 02:10:15 +03:00
|
|
|
char *prefix, char *hostid,
|
2008-02-28 04:57:57 +03:00
|
|
|
char *batchid, char *job, char *proc)
|
2006-06-29 01:03:31 +04:00
|
|
|
{
|
|
|
|
char *fulldirpath = NULL,
|
|
|
|
*frontend = NULL,
|
|
|
|
*sav = NULL;
|
|
|
|
int return_code = ORTE_SUCCESS, rtn;
|
|
|
|
/* This indicates if the prefix was set, and so if it fails then we
|
|
|
|
* should try with the default prefixes.*/
|
|
|
|
bool dbl_check_prefix = false;
|
2006-07-04 02:23:07 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
if( NULL != prefix)
|
|
|
|
dbl_check_prefix = true;
|
|
|
|
|
|
|
|
try_again:
|
|
|
|
/*
|
|
|
|
* If the first attempt at the path creation failed, try with a null
|
|
|
|
* prefix. unless the original prefix was null, then we fail.
|
|
|
|
*/
|
|
|
|
if(!dbl_check_prefix && /* an indicator that we are trying a second time */
|
|
|
|
NULL != prefix) {
|
|
|
|
free(prefix);
|
|
|
|
prefix = NULL;
|
2005-03-19 02:58:36 +03:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Get the session directory full name
|
|
|
|
* First try it with the specified prefix.
|
|
|
|
*/
|
|
|
|
if( ORTE_SUCCESS != ( rtn = orte_session_dir_get_name(&fulldirpath,
|
|
|
|
&prefix,
|
|
|
|
&frontend,
|
2008-03-24 02:10:15 +03:00
|
|
|
hostid,
|
2008-02-28 04:57:57 +03:00
|
|
|
batchid, job,
|
2006-06-29 01:03:31 +04:00
|
|
|
proc) ) ) {
|
|
|
|
return_code = rtn;
|
|
|
|
/*
|
|
|
|
* If the first attempt at the path creation failed, try with a null
|
|
|
|
* prefix. unless the original prefix was null, then we fail :(
|
|
|
|
*/
|
|
|
|
if(dbl_check_prefix) {
|
|
|
|
dbl_check_prefix = false;
|
|
|
|
goto try_again;
|
|
|
|
}
|
|
|
|
else {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(return_code);
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Now that we have the full path, go ahead and create it if necessary
|
|
|
|
*/
|
|
|
|
if( create ) {
|
|
|
|
if( ORTE_SUCCESS != (rtn = orte_create_dir(fulldirpath) ) ) {
|
|
|
|
return_code = rtn;
|
|
|
|
|
|
|
|
if(dbl_check_prefix) {
|
|
|
|
dbl_check_prefix = false;
|
|
|
|
goto try_again;
|
|
|
|
}
|
|
|
|
else {
|
2007-07-18 23:50:54 +04:00
|
|
|
ORTE_ERROR_LOG(return_code);
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* if we are not creating, then just verify that the path is OK
|
|
|
|
*/
|
|
|
|
else {
|
2006-07-04 02:23:07 +04:00
|
|
|
if( ORTE_SUCCESS != (rtn = opal_os_dirpath_access(fulldirpath, 0) )) {
|
2006-06-29 01:03:31 +04:00
|
|
|
/* It is not valid so we give up and return an error */
|
|
|
|
return_code = rtn;
|
|
|
|
|
|
|
|
if(dbl_check_prefix) {
|
|
|
|
dbl_check_prefix = false;
|
|
|
|
goto try_again;
|
|
|
|
}
|
|
|
|
else {
|
These changes were mostly captured in a prior RFC (except for #2 below) and are aimed specifically at improving startup performance and setting up the remaining modifications described in that RFC.
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component.
This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done:
As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in.
In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in.
The incoming changes revamp these procedures in three ways:
1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step.
The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic.
Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure.
2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed.
The size of this data has been reduced in three ways:
(a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes.
To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose.
(b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction.
(c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using.
While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly.
3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup.
It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging.
Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future.
There are a few minor additional changes in the commit that I'll just note in passing:
* propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details.
* requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details.
* cleanup of some stale header files
This commit was SVN r16364.
2007-10-05 23:48:23 +04:00
|
|
|
/* it is okay for the path not to be found - don't error
|
|
|
|
* log that case, but do error log others
|
|
|
|
*/
|
|
|
|
if (ORTE_ERR_NOT_FOUND != return_code) {
|
|
|
|
ORTE_ERROR_LOG(return_code);
|
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
goto cleanup;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2005-03-19 02:58:36 +03:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
return_code = ORTE_SUCCESS;
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* If we are creating the directory tree, the overwrite the
|
|
|
|
* global structure fields
|
|
|
|
*/
|
|
|
|
if (create) {
|
2005-03-19 02:58:36 +03:00
|
|
|
if (NULL != orte_process_info.tmpdir_base) {
|
|
|
|
free(orte_process_info.tmpdir_base);
|
|
|
|
orte_process_info.tmpdir_base = NULL;
|
|
|
|
}
|
|
|
|
if (NULL != orte_process_info.top_session_dir) {
|
|
|
|
free(orte_process_info.top_session_dir);
|
|
|
|
orte_process_info.top_session_dir = NULL;
|
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Update some of the global structures if they are empty
|
|
|
|
*/
|
|
|
|
if (NULL == orte_process_info.tmpdir_base)
|
|
|
|
orte_process_info.tmpdir_base = strdup(prefix);
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
if (NULL == orte_process_info.top_session_dir)
|
|
|
|
orte_process_info.top_session_dir = strdup(frontend);
|
|
|
|
|
2004-10-15 20:31:54 +04:00
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Set the process session directory
|
|
|
|
*/
|
2004-10-15 20:31:54 +04:00
|
|
|
if (NULL != proc) {
|
2005-03-19 02:58:36 +03:00
|
|
|
if (create) { /* overwrite if creating */
|
|
|
|
if (NULL != orte_process_info.proc_session_dir) {
|
2006-06-29 01:03:31 +04:00
|
|
|
free(orte_process_info.proc_session_dir);
|
|
|
|
orte_process_info.proc_session_dir = NULL;
|
2005-03-19 02:58:36 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (NULL == orte_process_info.proc_session_dir) {
|
|
|
|
orte_process_info.proc_session_dir = strdup(fulldirpath);
|
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/* Strip off last part of directory structure */
|
2006-08-23 07:32:36 +04:00
|
|
|
sav = opal_dirname(fulldirpath);
|
2006-06-29 01:03:31 +04:00
|
|
|
free(fulldirpath);
|
2006-08-23 07:32:36 +04:00
|
|
|
fulldirpath = sav;
|
2006-06-29 01:03:31 +04:00
|
|
|
sav = NULL;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
/*
|
|
|
|
* Set the job session directory
|
|
|
|
*/
|
2004-10-15 20:31:54 +04:00
|
|
|
if (NULL != job) {
|
2005-03-19 02:58:36 +03:00
|
|
|
if (create) { /* overwrite if creating */
|
|
|
|
if (NULL != orte_process_info.job_session_dir) {
|
2006-06-29 01:03:31 +04:00
|
|
|
free(orte_process_info.job_session_dir);
|
|
|
|
orte_process_info.job_session_dir = NULL;
|
2005-03-19 02:58:36 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
if (NULL == orte_process_info.job_session_dir) {
|
|
|
|
orte_process_info.job_session_dir = strdup(fulldirpath);
|
|
|
|
}
|
2006-06-29 01:03:31 +04:00
|
|
|
|
|
|
|
/* Strip off last part of directory structure */
|
2006-08-23 07:32:36 +04:00
|
|
|
sav = opal_dirname(fulldirpath);
|
2006-06-29 01:03:31 +04:00
|
|
|
free(fulldirpath);
|
2006-08-23 07:32:36 +04:00
|
|
|
fulldirpath = sav;
|
2006-06-29 01:03:31 +04:00
|
|
|
sav = NULL;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2005-03-14 23:57:21 +03:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "procdir: %s",
|
2005-06-27 03:11:37 +04:00
|
|
|
OMPI_PRINTF_FIX_STRING(orte_process_info.proc_session_dir));
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "jobdir: %s",
|
2005-06-27 03:11:37 +04:00
|
|
|
OMPI_PRINTF_FIX_STRING(orte_process_info.job_session_dir));
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "top: %s",
|
2005-06-27 03:11:37 +04:00
|
|
|
OMPI_PRINTF_FIX_STRING(orte_process_info.top_session_dir));
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "tmp: %s",
|
2005-06-27 03:11:37 +04:00
|
|
|
OMPI_PRINTF_FIX_STRING(orte_process_info.tmpdir_base));
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-06-29 01:03:31 +04:00
|
|
|
cleanup:
|
|
|
|
if(NULL != fulldirpath)
|
2004-10-15 20:31:54 +04:00
|
|
|
free(fulldirpath);
|
2006-06-29 01:03:31 +04:00
|
|
|
if(NULL != frontend)
|
|
|
|
free(frontend);
|
|
|
|
if(NULL != sav)
|
|
|
|
free(sav);
|
|
|
|
|
2004-10-15 20:31:54 +04:00
|
|
|
return return_code;
|
|
|
|
}
|
|
|
|
|
2006-02-16 03:16:22 +03:00
|
|
|
/*
|
2006-06-29 01:03:31 +04:00
|
|
|
* A job has aborted - so force cleanup of the session directory
|
2006-02-16 03:16:22 +03:00
|
|
|
*/
|
|
|
|
int
|
|
|
|
orte_session_dir_cleanup(orte_jobid_t jobid)
|
|
|
|
{
|
|
|
|
int rc;
|
|
|
|
char *tmp;
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
char *job=NULL, *job_session_dir=NULL;
|
2006-02-16 03:16:22 +03:00
|
|
|
|
|
|
|
/* need to setup the top_session_dir with the prefix */
|
|
|
|
tmp = opal_os_path(false,
|
|
|
|
orte_process_info.tmpdir_base,
|
|
|
|
orte_process_info.top_session_dir, NULL);
|
|
|
|
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
if (ORTE_JOBID_WILDCARD != jobid) {
|
|
|
|
/* define the proc and job session directories for this process */
|
2008-02-28 04:57:57 +03:00
|
|
|
if (ORTE_SUCCESS != (rc = orte_util_convert_jobid_to_string(&job, jobid))) {
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
free(tmp);
|
|
|
|
return rc;
|
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
job_session_dir = opal_os_path( false, orte_process_info.top_session_dir,
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
job, NULL );
|
|
|
|
if( NULL == job_session_dir ) {
|
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
|
|
|
|
free(tmp);
|
|
|
|
free(job);
|
|
|
|
return ORTE_ERR_OUT_OF_RESOURCE;
|
|
|
|
}
|
|
|
|
|
|
|
|
opal_os_dirpath_destroy(job_session_dir,
|
|
|
|
true, orte_dir_check_file);
|
|
|
|
} else {
|
|
|
|
/* if we want the session_dir removed for ALL jobids, then
|
2008-02-28 04:57:57 +03:00
|
|
|
* just recursively blow the whole session away, saving only
|
|
|
|
* output files
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
*/
|
2008-02-28 04:57:57 +03:00
|
|
|
opal_os_dirpath_destroy(tmp, true, orte_dir_check_file_output);
|
2006-02-16 03:16:22 +03:00
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
opal_os_dirpath_destroy(tmp,
|
|
|
|
false, orte_dir_check_file);
|
|
|
|
|
This commit represents a bunch of work on a Mercurial side branch. As
such, the commit message back to the master SVN repository is fairly
long.
= ORTE Job-Level Output Messages =
Add two new interfaces that should be used for all new code throughout
the ORTE and OMPI layers (we already make the search-and-replace on
the existing ORTE / OMPI layers):
* orte_output(): (and corresponding friends ORTE_OUTPUT,
orte_output_verbose, etc.) This function sends the output directly
to the HNP for processing as part of a job-specific output
channel. It supports all the same outputs as opal_output()
(syslog, file, stdout, stderr), but for stdout/stderr, the output
is sent to the HNP for processing and output. More on this below.
* orte_show_help(): This function is a drop-in-replacement for
opal_show_help(), with two differences in functionality:
1. the rendered text help message output is sent to the HNP for
display (rather than outputting directly into the process' stderr
stream)
1. the HNP detects duplicate help messages and does not display them
(so that you don't see the same error message N times, once from
each of your N MPI processes); instead, it counts "new" instances
of the help message and displays a message every ~5 seconds when
there are new ones ("I got X new copies of the help message...")
opal_show_help and opal_output still exist, but they only output in
the current process. The intent for the new orte_* functions is that
they can apply job-level intelligence to the output. As such, we
recommend that all new ORTE and OMPI code use the new orte_*
functions, not thei opal_* functions.
=== New code ===
For ORTE and OMPI programmers, here's what you need to do differently
in new code:
* Do not include opal/util/show_help.h or opal/util/output.h.
Instead, include orte/util/output.h (this one header file has
declarations for both the orte_output() series of functions and
orte_show_help()).
* Effectively s/opal_output/orte_output/gi throughout your code.
Note that orte_output_open() takes a slightly different argument
list (as a way to pass data to the filtering stream -- see below),
so you if explicitly call opal_output_open(), you'll need to
slightly adapt to the new signature of orte_output_open().
* Literally s/opal_show_help/orte_show_help/. The function signature
is identical.
=== Notes ===
* orte_output'ing to stream 0 will do similar to what
opal_output'ing did, so leaving a hard-coded "0" as the first
argument is safe.
* For systems that do not use ORTE's RML or the HNP, the effect of
orte_output_* and orte_show_help will be identical to their opal
counterparts (the additional information passed to
orte_output_open() will be lost!). Indeed, the orte_* functions
simply become trivial wrappers to their opal_* counterparts. Note
that we have not tested this; the code is simple but it is quite
possible that we mucked something up.
= Filter Framework =
Messages sent view the new orte_* functions described above and
messages output via the IOF on the HNP will now optionally be passed
through a new "filter" framework before being output to
stdout/stderr. The "filter" OPAL MCA framework is intended to allow
preprocessing to messages before they are sent to their final
destinations. The first component that was written in the filter
framework was to create an XML stream, segregating all the messages
into different XML tags, etc. This will allow 3rd party tools to read
the stdout/stderr from the HNP and be able to know exactly what each
text message is (e.g., a help message, another OMPI infrastructure
message, stdout from the user process, stderr from the user process,
etc.).
Filtering is not active by default. Filter components must be
specifically requested, such as:
{{{
$ mpirun --mca filter xml ...
}}}
There can only be one filter component active.
= New MCA Parameters =
The new functionality described above introduces two new MCA
parameters:
* '''orte_base_help_aggregate''': Defaults to 1 (true), meaning that
help messages will be aggregated, as described above. If set to 0,
all help messages will be displayed, even if they are duplicates
(i.e., the original behavior).
* '''orte_base_show_output_recursions''': An MCA parameter to help
debug one of the known issues, described below. It is likely that
this MCA parameter will disappear before v1.3 final.
= Known Issues =
* The XML filter component is not complete. The current output from
this component is preliminary and not real XML. A bit more work
needs to be done to configure.m4 search for an appropriate XML
library/link it in/use it at run time.
* There are possible recursion loops in the orte_output() and
orte_show_help() functions -- e.g., if RML send calls orte_output()
or orte_show_help(). We have some ideas how to fix these, but
figured that it was ok to commit before feature freeze with known
issues. The code currently contains sub-optimal workarounds so
that this will not be a problem, but it would be good to actually
solve the problem rather than have hackish workarounds before v1.3 final.
This commit was SVN r18434.
2008-05-14 00:00:55 +04:00
|
|
|
if (NULL != job_session_dir && opal_os_dirpath_is_empty(job_session_dir)) {
|
2006-02-16 03:16:22 +03:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found job session dir empty - deleting");
|
2006-02-16 03:16:22 +03:00
|
|
|
}
|
|
|
|
rmdir(job_session_dir);
|
|
|
|
} else {
|
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: job session dir not empty - leaving");
|
2006-02-16 03:16:22 +03:00
|
|
|
}
|
|
|
|
goto CLEANUP;
|
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
if (opal_os_dirpath_is_empty(tmp)) {
|
2006-02-16 03:16:22 +03:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found top session dir empty - deleting");
|
2006-02-16 03:16:22 +03:00
|
|
|
}
|
|
|
|
rmdir(tmp);
|
|
|
|
} else {
|
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: top session dir not empty - leaving");
|
2006-02-16 03:16:22 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
CLEANUP:
|
|
|
|
free(tmp);
|
Bring in the generalized xcast communication system along with the correspondingly revised orted launch. I will send a message out to developers explaining the basic changes. In brief:
1. generalize orte_rml.xcast to become a general broadcast-like messaging system. Messages can now be sent to any tag on the daemons or processes. Note that any message sent via xcast will be delivered to ALL processes in the specified job - you don't get to pick and choose. At a later date, we will introduce an augmented capability that will use the daemons as relays, but will allow you to send to a specified array of process names.
2. extended orte_rml.xcast so it supports more scalable message routing methodologies. At the moment, we support three: (a) direct, which sends the message directly to all recipients; (b) linear, which sends the message to the local daemon on each node, which then relays it to its own local procs; and (b) binomial, which sends the message via a binomial algo across all the daemons, each of which then relays to its own local procs. The crossover points between the algos are adjustable via MCA param, or you can simply demand that a specific algo be used.
3. orteds no longer exhibit two types of behavior: bootproxy or VM. Orteds now always behave like they are part of a virtual machine - they simply launch a job if mpirun tells them to do so. This is another step towards creating an "orteboot" functionality, but also provided a clean system for supporting message relaying.
Note one major impact of this commit: multiple daemons on a node cannot be supported any longer! Only a single daemon/node is now allowed.
This commit is known to break support for the following environments: POE, Xgrid, Xcpu, Windows. It has been tested on rsh, SLURM, and Bproc. Modifications for TM support have been made but could not be verified due to machine problems at LANL. Modifications for SGE have been made but could not be verified. The developers for the non-verified environments will be separately notified along with suggestions on how to fix the problems.
This commit was SVN r15007.
2007-06-12 17:28:54 +04:00
|
|
|
if (NULL != job) free(job);
|
|
|
|
if (NULL != job_session_dir) free(job_session_dir);
|
2006-02-16 03:16:22 +03:00
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
2004-10-15 20:31:54 +04:00
|
|
|
|
|
|
|
int
|
2005-04-14 05:04:26 +04:00
|
|
|
orte_session_dir_finalize(orte_process_name_t *proc)
|
2004-10-15 20:31:54 +04:00
|
|
|
{
|
2005-04-14 05:04:26 +04:00
|
|
|
int rc;
|
2005-04-07 23:19:48 +04:00
|
|
|
char *tmp;
|
2005-04-14 05:04:26 +04:00
|
|
|
char *job, *job_session_dir, *vpid, *proc_session_dir;
|
2005-05-13 01:44:23 +04:00
|
|
|
|
2005-04-07 23:19:48 +04:00
|
|
|
/* need to setup the top_session_dir with the prefix */
|
2005-08-30 01:35:01 +04:00
|
|
|
tmp = opal_os_path(false,
|
|
|
|
orte_process_info.tmpdir_base,
|
|
|
|
orte_process_info.top_session_dir, NULL);
|
2005-04-07 23:19:48 +04:00
|
|
|
|
2005-04-14 05:04:26 +04:00
|
|
|
/* define the proc and job session directories for this process */
|
2008-02-28 04:57:57 +03:00
|
|
|
if (ORTE_SUCCESS != (rc = orte_util_convert_jobid_to_string(&job, proc->jobid))) {
|
2005-04-14 05:04:26 +04:00
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
free(tmp);
|
|
|
|
return rc;
|
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
if (ORTE_SUCCESS != (rc = orte_util_convert_vpid_to_string(&vpid, proc->vpid))) {
|
2005-04-14 05:04:26 +04:00
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
free(tmp);
|
|
|
|
free(job);
|
|
|
|
return rc;
|
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
job_session_dir = opal_os_path( false, orte_process_info.top_session_dir,
|
2006-08-23 07:32:36 +04:00
|
|
|
job, NULL );
|
|
|
|
if( NULL == job_session_dir ) {
|
2005-04-14 05:04:26 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
|
|
|
|
free(tmp);
|
|
|
|
free(job);
|
|
|
|
free(vpid);
|
|
|
|
return ORTE_ERR_OUT_OF_RESOURCE;
|
|
|
|
}
|
2006-08-23 07:32:36 +04:00
|
|
|
proc_session_dir = opal_os_path( false, job_session_dir, vpid, NULL );
|
|
|
|
if( NULL == proc_session_dir ) {
|
2005-04-14 05:04:26 +04:00
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_OUT_OF_RESOURCE);
|
|
|
|
free(tmp);
|
|
|
|
free(job);
|
|
|
|
free(vpid);
|
|
|
|
free(job_session_dir);
|
|
|
|
return ORTE_ERR_OUT_OF_RESOURCE;
|
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
opal_os_dirpath_destroy(proc_session_dir,
|
|
|
|
false, orte_dir_check_file);
|
|
|
|
opal_os_dirpath_destroy(job_session_dir,
|
|
|
|
false, orte_dir_check_file);
|
2008-02-28 04:57:57 +03:00
|
|
|
opal_os_dirpath_destroy(orte_process_info.top_session_dir,
|
2006-07-04 02:23:07 +04:00
|
|
|
false, orte_dir_check_file);
|
|
|
|
opal_os_dirpath_destroy(tmp,
|
|
|
|
false, orte_dir_check_file);
|
|
|
|
|
|
|
|
if (opal_os_dirpath_is_empty(proc_session_dir)) {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found proc session dir empty - deleting");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2005-04-14 05:04:26 +04:00
|
|
|
rmdir(proc_session_dir);
|
2004-10-15 20:31:54 +04:00
|
|
|
} else {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: proc session dir not empty - leaving");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2005-04-14 05:04:26 +04:00
|
|
|
goto CLEANUP;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
if (opal_os_dirpath_is_empty(job_session_dir)) {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found job session dir empty - deleting");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2005-04-14 05:04:26 +04:00
|
|
|
rmdir(job_session_dir);
|
2004-10-15 20:31:54 +04:00
|
|
|
} else {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: job session dir not empty - leaving");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2005-04-14 05:04:26 +04:00
|
|
|
goto CLEANUP;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2008-02-28 04:57:57 +03:00
|
|
|
if (opal_os_dirpath_is_empty(orte_process_info.top_session_dir)) {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found top session dir empty - deleting");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
rmdir(orte_process_info.top_session_dir);
|
2004-10-15 20:31:54 +04:00
|
|
|
} else {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: top session dir not empty - leaving");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2005-04-14 05:04:26 +04:00
|
|
|
goto CLEANUP;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
if (opal_os_dirpath_is_empty(tmp)) {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: found top session dir empty - deleting");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
|
|
|
rmdir(tmp);
|
2004-10-15 20:31:54 +04:00
|
|
|
} else {
|
2005-04-07 23:19:48 +04:00
|
|
|
if (orte_debug_flag) {
|
2008-06-09 18:53:58 +04:00
|
|
|
opal_output(0, "sess_dir_finalize: top session dir not empty - leaving");
|
2005-04-07 23:19:48 +04:00
|
|
|
}
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2005-04-14 05:04:26 +04:00
|
|
|
CLEANUP:
|
2005-04-07 23:19:48 +04:00
|
|
|
free(tmp);
|
2005-04-14 05:04:26 +04:00
|
|
|
free(job);
|
|
|
|
free(vpid);
|
|
|
|
free(job_session_dir);
|
|
|
|
free(proc_session_dir);
|
2005-08-15 22:25:35 +04:00
|
|
|
return ORTE_SUCCESS;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
|
|
|
|
2006-07-04 02:23:07 +04:00
|
|
|
static bool
|
2008-02-28 04:57:57 +03:00
|
|
|
orte_dir_check_file(const char *root, const char *path)
|
|
|
|
{
|
2006-07-04 02:23:07 +04:00
|
|
|
/*
|
|
|
|
* Keep:
|
|
|
|
* - files starting with "output-"
|
2008-02-28 04:57:57 +03:00
|
|
|
* - files that indicate abort
|
2006-07-04 02:23:07 +04:00
|
|
|
*/
|
|
|
|
if( (0 == strncmp(path, "output-", strlen("output-"))) ||
|
2008-02-28 04:57:57 +03:00
|
|
|
(0 == strcmp(path, "abort"))) {
|
2006-07-04 02:23:07 +04:00
|
|
|
return false;
|
2004-11-03 00:48:31 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
2004-10-15 20:31:54 +04:00
|
|
|
}
|
2008-02-28 04:57:57 +03:00
|
|
|
|
|
|
|
static bool
|
|
|
|
orte_dir_check_file_output(const char *root, const char *path)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Keep:
|
|
|
|
* - files starting with "output-"
|
|
|
|
*/
|
|
|
|
if( 0 == strncmp(path, "output-", strlen("output-"))) {
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|