Exemplo n.º 1
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    char **hosts = NULL;
    char *nodelist;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }
    
    /* Start by getting a unique name */
    tm_set_name();
    
    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        /* get the list of nodes used for this job */
        nodelist = getenv("OMPI_MCA_orte_nodelist");
        
        if (NULL != nodelist) {
            /* split the node list into an argv array */
            hosts = opal_argv_split(nodelist, ',');
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        opal_argv_free(hosts);
        return ORTE_SUCCESS;
    }
    
    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        return ORTE_SUCCESS;
        
    }
    
    /* no other options are supported! */
    error = "ess_error";
    ret = ORTE_ERROR;
    
error:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }

    return ret;
}
Exemplo n.º 2
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }

    /* Start by getting a unique name */
    slurm_set_name();

    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        return ORTE_SUCCESS;
    }

    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup(NULL))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        return ORTE_SUCCESS;

    }

    /* no other options are supported! */
    error = "ess_error";
    ret = ORTE_ERROR;

error:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }

    return ret;
}
Exemplo n.º 3
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    char **hosts = NULL;
    char *slurm_nodelist;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }
    
    /* Start by getting a unique name */
    slurm_set_name();
    
    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        /* get the list of nodes used for this job */
        mca_base_param_reg_string_name("orte", "nodelist", "List of nodes in job",
                                       true, false, NULL, &slurm_nodelist);
        
        if (NULL != slurm_nodelist) {
            /* split the node list into an argv array */
            hosts = opal_argv_split(slurm_nodelist, ',');
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        opal_argv_free(hosts);
        return ORTE_SUCCESS;
    }
    
    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        /* as a tool, I don't need a nidmap - so just return now */
        return ORTE_SUCCESS;
        
    }
    
    /* otherwise, I must be an application process - use
     * the default procedure to finish my setup
     */
    if (ORTE_SUCCESS != (ret = orte_ess_base_app_setup())) {
        ORTE_ERROR_LOG(ret);
        error = "orte_ess_base_app_setup";
        goto error;
    }
    /* setup the nidmap arrays */
    if (ORTE_SUCCESS != (ret = orte_util_nidmap_init(orte_process_info.sync_buf))) {
        ORTE_ERROR_LOG(ret);
        error = "orte_util_nidmap_init";
        goto error;
    }

    return ORTE_SUCCESS;
    
error:
    orte_show_help("help-orte-runtime.txt",
                   "orte_init:startup:internal-failure",
                   true, error, ORTE_ERROR_NAME(ret), ret);
    
    return ret;
}
Exemplo n.º 4
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    char **hosts = NULL;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }
    
    /* Start by getting a unique name from the enviro */
    env_set_name();

    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        if (NULL != orte_node_regex) {
            /* extract the nodes */
            if (ORTE_SUCCESS != (ret = orte_regex_extract_node_names(orte_node_regex, &hosts))) {
                error = "orte_regex_extract_node_names";
                goto error;
            }
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        opal_argv_free(hosts);
        return ORTE_SUCCESS;
    }
    
    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        /* as a tool, I don't need a nidmap - so just return now */
        return ORTE_SUCCESS;
        
    }
    
    /* otherwise, I must be an application process - use
     * the default procedure to finish my setup
     */
    if (ORTE_SUCCESS != (ret = orte_ess_base_app_setup())) {
        ORTE_ERROR_LOG(ret);
        error = "orte_ess_base_app_setup";
        goto error;
    }
    
    /* if data was provided, update the database */
    if (ORTE_SUCCESS != (ret = orte_util_nidmap_init(orte_process_info.sync_buf))) {
        ORTE_ERROR_LOG(ret);
        error = "orte_util_nidmap_init";
        goto error;
    }
    
    /* setup process binding */
    if (ORTE_SUCCESS != (ret = orte_ess_base_proc_binding())) {
        error = "proc_binding";
        goto error;
    }

    return ORTE_SUCCESS;

error:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }

    return ret;
}
Exemplo n.º 5
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    orte_jobid_t jobid;
    orte_vpid_t vpid;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }


    if (NULL != orte_ess_base_jobid &&
        NULL != orte_ess_base_vpid) {
        opal_output_verbose(2, orte_ess_base_framework.framework_output,
                            "ess:tool:obtaining name from environment");
        if (ORTE_SUCCESS != (ret = orte_util_convert_string_to_jobid(&jobid, orte_ess_base_jobid))) {
            return(ret);
        }
        ORTE_PROC_MY_NAME->jobid = jobid;
        if (ORTE_SUCCESS != (ret = orte_util_convert_string_to_vpid(&vpid, orte_ess_base_vpid))) {
            return(ret);
        }
        ORTE_PROC_MY_NAME->vpid = vpid;
    } else {
        /* If we are a tool with no name, then define it here */
        uint16_t jobfam;
        uint32_t hash32;
        uint32_t bias;

        opal_output_verbose(2, orte_ess_base_framework.framework_output,
                            "ess:tool:computing name");
        /* hash the nodename */
        OPAL_HASH_STR(orte_process_info.nodename, hash32);
        bias = (uint32_t)orte_process_info.pid;
        /* fold in the bias */
        hash32 = hash32 ^ bias;

        /* now compress to 16-bits */
        jobfam = (uint16_t)(((0x0000ffff & (0xffff0000 & hash32) >> 16)) ^ (0x0000ffff & hash32));

        /* set the name */
        ORTE_PROC_MY_NAME->jobid = 0xffff0000 & ((uint32_t)jobfam << 16);
        ORTE_PROC_MY_NAME->vpid = 0;
    }

    /* do the rest of the standard tool init */
    if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
        ORTE_ERROR_LOG(ret);
        error = "orte_ess_base_tool_setup";
        goto error;
    }

    return ORTE_SUCCESS;

 error:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }

    return ret;
}
Exemplo n.º 6
0
static int rte_init(char flags)
{
    int ret;
    char *error = NULL;
    orte_jmap_t *jmap;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }
    
    /* Start by getting a unique name */
    slurm_set_name();
    
    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (orte_process_info.daemon) {
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
    } else if (orte_process_info.tool) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
    } else {
        /* otherwise, I must be an application process - use
         * the default procedure to finish my setup
         */
        if (ORTE_SUCCESS != (ret = orte_ess_base_app_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_app_setup";
            goto error;
        }
        
        /* setup the nidmap arrays */
        OBJ_CONSTRUCT(&nidmap, opal_pointer_array_t);
        opal_pointer_array_init(&nidmap, 8, INT32_MAX, 8);
        
        /* setup array of jmaps */
        OBJ_CONSTRUCT(&jobmap, opal_pointer_array_t);
        opal_pointer_array_init(&jobmap, 1, INT32_MAX, 1);
        jmap = OBJ_NEW(orte_jmap_t);
        jmap->job = ORTE_PROC_MY_NAME->jobid;
        opal_pointer_array_add(&jobmap, jmap);
        
        /* if one was provided, build my nidmap */
        if (ORTE_SUCCESS != (ret = orte_ess_base_build_nidmap(orte_process_info.sync_buf,
                                                              &nidmap, &jmap->pmap, &nprocs))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_build_nidmap";
            goto error;
        }
    }
    
    return ORTE_SUCCESS;
    
error:
    orte_show_help("help-orte-runtime.txt",
                   "orte_init:startup:internal-failure",
                   true, error, ORTE_ERROR_NAME(ret), ret);
    
    return ret;
}
Exemplo n.º 7
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    char **hosts = NULL;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }
    
    /* Start by getting a unique name from the enviro */
    env_set_name();

    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        if (NULL != orte_node_regex) {
            /* extract the nodes */
            if (ORTE_SUCCESS != (ret = orte_regex_extract_node_names(orte_node_regex, &hosts))) {
                error = "orte_regex_extract_node_names";
                goto error;
            }
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        opal_argv_free(hosts);
        return ORTE_SUCCESS;
    }
    
    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        /* as a tool, I don't need a nidmap - so just return now */
        return ORTE_SUCCESS;
        
    }
    
    /* use the default procedure to finish my setup */
    if (ORTE_SUCCESS != (ret = orte_ess_base_app_setup(true))) {
        ORTE_ERROR_LOG(ret);
        error = "orte_ess_base_app_setup";
        goto error;
    }
    
    /* if data was provided, update the database */
    if (ORTE_SUCCESS != (ret = orte_util_nidmap_init(orte_process_info.sync_buf))) {
        ORTE_ERROR_LOG(ret);
        error = "orte_util_nidmap_init";
        goto error;
    }
    
    /* setup process binding */
    if (ORTE_SUCCESS != (ret = orte_ess_base_proc_binding())) {
        error = "proc_binding";
        goto error;
    }

    /* if we are an ORTE app - and not an MPI app - then
     * we need to exchange our connection info here.
     * MPI_Init has its own modex, so we don't need to do
     * two of them. However, if we don't do a modex at all,
     * then processes have no way to communicate
     *
     * NOTE: only do this when the process originally launches.
     * Cannot do this on a restart as the rest of the processes
     * in the job won't be executing this step, so we would hang
     */
    if (ORTE_PROC_IS_NON_MPI && !orte_do_not_barrier) {
        orte_grpcomm_collective_t coll;
        OBJ_CONSTRUCT(&coll, orte_grpcomm_collective_t);
        coll.id = orte_process_info.peer_modex;
        coll.active = true;
        if (ORTE_SUCCESS != (ret = orte_grpcomm.modex(&coll))) {
            ORTE_ERROR_LOG(ret);
            error = "orte modex";
            goto error;
        }
        ORTE_WAIT_FOR_COMPLETION(coll.active);
        OBJ_DESTRUCT(&coll);
    }
    
    return ORTE_SUCCESS;

error:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }

    return ret;
}
Exemplo n.º 8
0
static int rte_init(void)
{
    int ret;
    char *error = NULL;
    char **hosts = NULL;

    OPAL_OUTPUT_VERBOSE((1, orte_ess_base_framework.framework_output,
                         "ess:alps in rte_init"));

    /*
     * shouldn't have been able to open this ess component if
     * process is app proc
     */

    if (ORTE_PROC_IS_APP) {
        error = "mpi rank invoking alps rte_init";
        ret = ORTE_ERR_NOT_SUPPORTED;
        goto fn_fail;
    }

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto fn_fail;
    }

    if (ORTE_SUCCESS != (ret = alps_set_name())) {
        error = "alps_set_name";
        goto fn_fail;
    }

    /*
     * if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        if (NULL != orte_node_regex) {
            /* extract the nodes */
            if (ORTE_SUCCESS != (ret =
                orte_regex_extract_node_names(orte_node_regex, &hosts)) ||
                NULL == hosts) {
                error = "orte_regex_extract_node_names";
                goto fn_fail;
            }
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto fn_fail;
        }
        if (NULL != hosts) {
            opal_argv_free(hosts);
        }

        /*
         * now synchronize with aprun.
         */

        if (ORTE_SUCCESS != (ret = orte_ess_alps_sync_start())) {
            error = "orte_ess_alps_sync";
            goto fn_fail;
        }

        ret = ORTE_SUCCESS;
        goto fn_exit;
    }

    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto fn_fail;
        }
        /* as a tool, I don't need a nidmap - so just return now */
        ret = ORTE_SUCCESS;
        goto fn_exit;
    }

   fn_exit:
    return ret;

   fn_fail:
    if (ORTE_ERR_SILENT != ret && !orte_report_silent_errors) {
        orte_show_help("help-orte-runtime.txt",
                       "orte_init:startup:internal-failure",
                       true, error, ORTE_ERROR_NAME(ret), ret);
    }
    goto fn_exit;
}
Exemplo n.º 9
0
static int rte_init(void)
{
    int ret, i;
    char *error = NULL;
    char **hosts = NULL;

    /* run the prolog */
    if (ORTE_SUCCESS != (ret = orte_ess_base_std_prolog())) {
        error = "orte_ess_base_std_prolog";
        goto error;
    }

    /* Start by getting a unique name */
    alps_set_name();

    /* if I am a daemon, complete my setup using the
     * default procedure
     */
    if (ORTE_PROC_IS_DAEMON) {
        if (NULL != orte_node_regex) {
            /* extract the nodes */
            if (ORTE_SUCCESS != (ret =
                orte_regex_extract_node_names(orte_node_regex, &hosts)) ||
                NULL == hosts) {
                error = "orte_regex_extract_node_names";
                goto error;
            }

            /* find our host in the list */
            for (i=0; NULL != hosts[i]; i++) {
                if (0 == strncmp(hosts[i], orte_process_info.nodename,
                                 strlen(hosts[i]))) {
                    /* correct our vpid */
                    ORTE_PROC_MY_NAME->vpid = starting_vpid + i;
                    OPAL_OUTPUT_VERBOSE((1, orte_ess_base_output,
                                         "ess:alps reset name to %s",
                                         ORTE_NAME_PRINT(ORTE_PROC_MY_NAME)));
                    break;
                }
            }
        }
        if (ORTE_SUCCESS != (ret = orte_ess_base_orted_setup(hosts))) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_orted_setup";
            goto error;
        }
        opal_argv_free(hosts);
        return ORTE_SUCCESS;
    }
    if (ORTE_PROC_IS_TOOL) {
        /* otherwise, if I am a tool proc, use that procedure */
        if (ORTE_SUCCESS != (ret = orte_ess_base_tool_setup())) {
            ORTE_ERROR_LOG(ret);
            error = "orte_ess_base_tool_setup";
            goto error;
        }
        /* as a tool, I don't need a nidmap - so just return now */
        return ORTE_SUCCESS;
    }
    /* otherwise, I must be an application process - use
     * the default procedure to finish my setup
     */
    if (ORTE_SUCCESS != (ret = orte_ess_base_app_setup())) {
        ORTE_ERROR_LOG(ret);
        error = "orte_ess_base_app_setup";
        goto error;
    }
    /* setup the nidmap arrays */
    if (ORTE_SUCCESS !=
        (ret = orte_util_nidmap_init(orte_process_info.sync_buf))) {
        ORTE_ERROR_LOG(ret);
        error = "orte_util_nidmap_init";
        goto error;
    }

    return ORTE_SUCCESS;

error:
    orte_show_help("help-orte-runtime.txt",
                   "orte_init:startup:internal-failure",
                   true, error, ORTE_ERROR_NAME(ret), ret);
    return ret;
}