2008-02-28 01:57:57 +00:00
|
|
|
/*
|
A number of C/R enhancements per RFC below:
http://www.open-mpi.org/community/lists/devel/2010/07/8240.php
Documentation:
http://osl.iu.edu/research/ft/
Major Changes:
--------------
* Added C/R-enabled Debugging support.
Enabled with the --enable-crdebug flag. See the following website for more information:
http://osl.iu.edu/research/ft/crdebug/
* Added Stable Storage (SStore) framework for checkpoint storage
* 'central' component does a direct to central storage save
* 'stage' component stages checkpoints to central storage while the application continues execution.
* 'stage' supports offline compression of checkpoints before moving (sstore_stage_compress)
* 'stage' supports local caching of checkpoints to improve automatic recovery (sstore_stage_caching)
* Added Compression (compress) framework to support
* Add two new ErrMgr recovery policies
* {{{crmig}}} C/R Process Migration
* {{{autor}}} C/R Automatic Recovery
* Added the {{{ompi-migrate}}} command line tool to support the {{{crmig}}} ErrMgr component
* Added CR MPI Ext functions (enable them with {{{--enable-mpi-ext=cr}}} configure option)
* {{{OMPI_CR_Checkpoint}}} (Fixes trac:2342)
* {{{OMPI_CR_Restart}}}
* {{{OMPI_CR_Migrate}}} (may need some more work for mapping rules)
* {{{OMPI_CR_INC_register_callback}}} (Fixes trac:2192)
* {{{OMPI_CR_Quiesce_start}}}
* {{{OMPI_CR_Quiesce_checkpoint}}}
* {{{OMPI_CR_Quiesce_end}}}
* {{{OMPI_CR_self_register_checkpoint_callback}}}
* {{{OMPI_CR_self_register_restart_callback}}}
* {{{OMPI_CR_self_register_continue_callback}}}
* The ErrMgr predicted_fault() interface has been changed to take an opal_list_t of ErrMgr defined types. This will allow us to better support a wider range of fault prediction services in the future.
* Add a progress meter to:
* FileM rsh (filem_rsh_process_meter)
* SnapC full (snapc_full_progress_meter)
* SStore stage (sstore_stage_progress_meter)
* Added 2 new command line options to ompi-restart
* --showme : Display the full command line that would have been exec'ed.
* --mpirun_opts : Command line options to pass directly to mpirun. (Fixes trac:2413)
* Deprecated some MCA params:
* crs_base_snapshot_dir deprecated, use sstore_stage_local_snapshot_dir
* snapc_base_global_snapshot_dir deprecated, use sstore_base_global_snapshot_dir
* snapc_base_global_shared deprecated, use sstore_stage_global_is_shared
* snapc_base_store_in_place deprecated, replaced with different components of SStore
* snapc_base_global_snapshot_ref deprecated, use sstore_base_global_snapshot_ref
* snapc_base_establish_global_snapshot_dir deprecated, never well supported
* snapc_full_skip_filem deprecated, use sstore_stage_skip_filem
Minor Changes:
--------------
* Fixes trac:1924 : {{{ompi-restart}}} now recognizes path prefixed checkpoint handles and does the right thing.
* Fixes trac:2097 : {{{ompi-info}}} should now report all available CRS components
* Fixes trac:2161 : Manual checkpoint movement. A user can 'mv' a checkpoint directory from the original location to another and still restart from it.
* Fixes trac:2208 : Honor various TMPDIR varaibles instead of forcing {{{/tmp}}}
* Move {{{ompi_cr_continue_like_restart}}} to {{{orte_cr_continue_like_restart}}} to be more flexible in where this should be set.
* opal_crs_base_metadata_write* functions have been moved to SStore to support a wider range of metadata handling functionality.
* Cleanup the CRS framework and components to work with the SStore framework.
* Cleanup the SnapC framework and components to work with the SStore framework (cleans up these code paths considerably).
* Add 'quiesce' hook to CRCP for a future enhancement.
* We now require a BLCR version that supports {{{cr_request_file()}}} or {{{cr_request_checkpoint()}}} in order to make the code more maintainable. Note that {{{cr_request_file}}} has been deprecated since 0.7.0, so we prefer to use {{{cr_request_checkpoint()}}}.
* Add optional application level INC callbacks (registered through the CR MPI Ext interface).
* Increase the {{{opal_cr_thread_sleep_wait}}} parameter to 1000 microseconds to make the C/R thread less aggressive.
* {{{opal-restart}}} now looks for cache directories before falling back on stable storage when asked.
* {{{opal-restart}}} also support local decompression before restarting
* {{{orte-checkpoint}}} now uses the SStore framework to work with the metadata
* {{{orte-restart}}} now uses the SStore framework to work with the metadata
* Remove the {{{orte-restart}}} preload option. This was removed since the user only needs to select the 'stage' component in order to support this functionality.
* Since the '-am' parameter is saved in the metadata, {{{ompi-restart}}} no longer hard codes {{{-am ft-enable-cr}}}.
* Fix {{{hnp}}} ErrMgr so that if a previous component in the stack has 'fixed' the problem, then it should be skipped.
* Make sure to decrement the number of 'num_local_procs' in the orted when one goes away.
* odls now checks the SStore framework to see if it needs to load any checkpoint files before launching (to support 'stage'). This separates the SStore logic from the --preload-[binary|files] options.
* Add unique IDs to the named pipes established between the orted and the app in SnapC. This is to better support migration and automatic recovery activities.
* Improve the checks for 'already checkpointing' error path.
* A a recovery output timer, to show how long it takes to restart a job
* Do a better job of cleaning up the old session directory on restart.
* Add a local module to the autor and crmig ErrMgr components. These small modules prevent the 'orted' component from attempting a local recovery (Which does not work for MPI apps at the moment)
* Add a fix for bounding the checkpointable region between MPI_Init and MPI_Finalize.
This commit was SVN r23587.
The following Trac tickets were found above:
Ticket 1924 --> https://svn.open-mpi.org/trac/ompi/ticket/1924
Ticket 2097 --> https://svn.open-mpi.org/trac/ompi/ticket/2097
Ticket 2161 --> https://svn.open-mpi.org/trac/ompi/ticket/2161
Ticket 2192 --> https://svn.open-mpi.org/trac/ompi/ticket/2192
Ticket 2208 --> https://svn.open-mpi.org/trac/ompi/ticket/2208
Ticket 2342 --> https://svn.open-mpi.org/trac/ompi/ticket/2342
Ticket 2413 --> https://svn.open-mpi.org/trac/ompi/ticket/2413
2010-08-10 20:51:11 +00:00
|
|
|
* Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana
|
2008-02-28 01:57:57 +00:00
|
|
|
* University Research and Technology
|
|
|
|
* Corporation. All rights reserved.
|
2011-06-23 20:38:02 +00:00
|
|
|
* Copyright (c) 2004-2011 The University of Tennessee and The University
|
2008-02-28 01:57:57 +00:00
|
|
|
* of Tennessee Research Foundation. All rights
|
|
|
|
* reserved.
|
|
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
|
|
* University of Stuttgart. All rights reserved.
|
|
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
|
|
* All rights reserved.
|
2010-09-15 20:17:50 +00:00
|
|
|
* Copyright (c) 2010 Oracle and/or its affiliates. All rights reserved.
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
* Copyright (c) 2011 Cisco Systems, Inc. All rights reserved.
|
2013-02-28 01:35:55 +00:00
|
|
|
* Copyright (c) 2011-2013 Los Alamos National Security, LLC.
|
2012-04-06 14:23:13 +00:00
|
|
|
* All rights reserved.
|
2008-02-28 01:57:57 +00:00
|
|
|
* $COPYRIGHT$
|
|
|
|
*
|
|
|
|
* Additional copyrights may follow
|
|
|
|
*
|
|
|
|
* $HEADER$
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "orte_config.h"
|
|
|
|
#include "orte/types.h"
|
|
|
|
|
|
|
|
#include <sys/types.h>
|
|
|
|
|
|
|
|
#include "opal/util/argv.h"
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
#include "opal/mca/hwloc/base/base.h"
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
#include "orte/mca/errmgr/errmgr.h"
|
2011-03-09 18:51:36 +00:00
|
|
|
#include "orte/mca/rmaps/base/base.h"
|
2008-02-28 01:57:57 +00:00
|
|
|
#include "opal/dss/dss.h"
|
|
|
|
#include "orte/util/name_fns.h"
|
2010-04-26 22:15:57 +00:00
|
|
|
#include "orte/util/error_strings.h"
|
2008-09-23 15:46:34 +00:00
|
|
|
#include "orte/runtime/orte_globals.h"
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
#include "orte/runtime/data_type_support/orte_dt_support.h"
|
|
|
|
|
|
|
|
static void orte_dt_quick_print(char **output, char *type_name, char *prefix, void *src, opal_data_type_t real_type)
|
|
|
|
{
|
|
|
|
int8_t *i8;
|
|
|
|
int16_t *i16;
|
|
|
|
int32_t *i32;
|
|
|
|
int64_t *i64;
|
|
|
|
uint8_t *ui8;
|
|
|
|
uint16_t *ui16;
|
|
|
|
uint32_t *ui32;
|
|
|
|
uint64_t *ui64;
|
|
|
|
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* check for NULL ptr */
|
|
|
|
if (NULL == src) {
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 8-bit\tValue: NULL pointer",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch(real_type) {
|
|
|
|
case OPAL_INT8:
|
|
|
|
i8 = (int8_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 8-bit\tValue: %d",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (int) *i8);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_UINT8:
|
|
|
|
ui8 = (uint8_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 8-bit\tValue: %u",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (unsigned int)*ui8);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_INT16:
|
|
|
|
i16 = (int16_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 16-bit\tValue: %d",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (int) *i16);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_UINT16:
|
|
|
|
ui16 = (uint16_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 16-bit\tValue: %u",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (unsigned int) *ui16);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_INT32:
|
|
|
|
i32 = (int32_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 32-bit\tValue: %ld",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (long) *i32);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_UINT32:
|
|
|
|
ui32 = (uint32_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 32-bit\tValue: %lu",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (unsigned long) *ui32);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_INT64:
|
|
|
|
i64 = (int64_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 64-bit\tValue: %ld",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (long) *i64);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case OPAL_UINT64:
|
|
|
|
ui64 = (uint64_t*)src;
|
|
|
|
asprintf(output, "%sData type: %s\tData size: 64-bit\tValue: %lu",
|
|
|
|
(NULL == prefix) ? "" : prefix, type_name, (unsigned long) *ui64);
|
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* STANDARD PRINT FUNCTION - WORKS FOR EVERYTHING NON-STRUCTURED
|
|
|
|
*/
|
|
|
|
int orte_dt_std_print(char **output, char *prefix, void *src, opal_data_type_t type)
|
|
|
|
{
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
switch(type) {
|
|
|
|
case ORTE_STD_CNTR:
|
|
|
|
orte_dt_quick_print(output, "ORTE_STD_CNTR", prefix, src, ORTE_STD_CNTR_T);
|
|
|
|
break;
|
2011-06-23 20:38:02 +00:00
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
case ORTE_VPID:
|
|
|
|
orte_dt_quick_print(output, "ORTE_VPID", prefix, src, ORTE_VPID_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_JOBID:
|
|
|
|
asprintf(output, "%sData Type: ORTE_JOBID\tData size: %lu\tValue: %s",
|
|
|
|
(NULL == prefix) ? "" : prefix, (unsigned long)sizeof(orte_jobid_t),
|
2009-03-05 21:50:47 +00:00
|
|
|
ORTE_JOBID_PRINT(*(orte_jobid_t*)src));
|
2008-02-28 01:57:57 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_PROC_STATE:
|
|
|
|
orte_dt_quick_print(output, "ORTE_PROC_STATE", prefix, src, ORTE_PROC_STATE_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_JOB_STATE:
|
|
|
|
orte_dt_quick_print(output, "ORTE_JOB_STATE", prefix, src, ORTE_JOB_STATE_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_NODE_STATE:
|
|
|
|
orte_dt_quick_print(output, "ORTE_NODE_STATE", prefix, src, ORTE_NODE_STATE_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_EXIT_CODE:
|
|
|
|
orte_dt_quick_print(output, "ORTE_EXIT_CODE", prefix, src, ORTE_EXIT_CODE_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_RML_TAG:
|
|
|
|
orte_dt_quick_print(output, "ORTE_RML_TAG", prefix, src, ORTE_RML_TAG_T);
|
|
|
|
break;
|
|
|
|
|
|
|
|
case ORTE_DAEMON_CMD:
|
|
|
|
orte_dt_quick_print(output, "ORTE_DAEMON_CMD", prefix, src, ORTE_DAEMON_CMD_T);
|
|
|
|
break;
|
|
|
|
|
Roll in the revamped IOF subsystem. Per the devel mailing list email, this is a complete rewrite of the iof framework designed to simplify the code for maintainability, and to support features we had planned to do, but were too difficult to implement in the old code. Specifically, the new code:
1. completely and cleanly separates responsibilities between the HNP, orted, and tool components.
2. removes all wireup messaging during launch and shutdown.
3. maintains flow control for stdin to avoid large-scale consumption of memory by orteds when large input files are forwarded. This is done using an xon/xoff protocol.
4. enables specification of stdin recipients on the mpirun cmd line. Allowed options include rank, "all", or "none". Default is rank 0.
5. creates a new MPI_Info key "ompi_stdin_target" that supports the above options for child jobs. Default is "none".
6. adds a new tool "orte-iof" that can connect to a running mpirun and display the output. Cmd line options allow selection of any combination of stdout, stderr, and stddiag. Default is stdout.
7. adds a new mpirun and orte-iof cmd line option "tag-output" that will tag each line of output with process name and stream ident. For example, "[1,0]<stdout>this is output"
This is not intended for the 1.3 release as it is a major change requiring considerable soak time.
This commit was SVN r19767.
2008-10-18 00:00:49 +00:00
|
|
|
case ORTE_IOF_TAG:
|
|
|
|
orte_dt_quick_print(output, "ORTE_IOF_TAG", prefix, src, ORTE_IOF_TAG_T);
|
|
|
|
break;
|
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
default:
|
|
|
|
ORTE_ERROR_LOG(ORTE_ERR_UNKNOWN_DATA_TYPE);
|
|
|
|
return ORTE_ERR_UNKNOWN_DATA_TYPE;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* NAME
|
|
|
|
*/
|
|
|
|
int orte_dt_print_name(char **output, char *prefix, orte_process_name_t *name, opal_data_type_t type)
|
|
|
|
{
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
if (NULL == name) {
|
|
|
|
asprintf(output, "%sData type: ORTE_PROCESS_NAME\tData Value: NULL",
|
|
|
|
(NULL == prefix ? " " : prefix));
|
|
|
|
} else {
|
|
|
|
asprintf(output, "%sData type: ORTE_PROCESS_NAME\tData Value: %s",
|
2009-03-05 21:50:47 +00:00
|
|
|
(NULL == prefix ? " " : prefix), ORTE_NAME_PRINT(name));
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* JOB
|
|
|
|
*/
|
|
|
|
int orte_dt_print_job(char **output, char *prefix, orte_job_t *src, opal_data_type_t type)
|
|
|
|
{
|
|
|
|
char *tmp, *tmp2, *tmp3, *pfx2, *pfx;
|
|
|
|
int32_t i;
|
|
|
|
int rc;
|
2009-07-13 14:49:20 +00:00
|
|
|
orte_app_context_t *app;
|
|
|
|
orte_proc_t *proc;
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* protect against NULL prefix */
|
|
|
|
if (NULL == prefix) {
|
|
|
|
asprintf(&pfx2, " ");
|
|
|
|
} else {
|
|
|
|
asprintf(&pfx2, "%s", prefix);
|
|
|
|
}
|
|
|
|
|
2012-09-11 20:35:46 +00:00
|
|
|
asprintf(&tmp, "\n%sData for job: %s\tRecovery: %s(%s)\n%s\tNum apps: %ld\tControls: %0x\tMPI allowed: %s\tStdin target: %s\tState: %s\tAbort: %s", pfx2,
|
2010-06-24 01:57:27 +00:00
|
|
|
ORTE_JOBID_PRINT(src->jobid),
|
2011-04-18 14:46:42 +00:00
|
|
|
(src->enable_recovery) ? "ENABLED" : "DISABLED",
|
|
|
|
(src->recovery_defined) ? "DEFINED" : "DEFAULT",
|
|
|
|
pfx2,
|
2012-09-11 20:35:46 +00:00
|
|
|
(long)src->num_apps, src->controls,
|
|
|
|
src->gang_launched ? "YES" : "NO", ORTE_VPID_PRINT(src->stdin_target),
|
2010-04-26 22:15:57 +00:00
|
|
|
orte_job_state_to_str(src->state), src->abort ? "True" : "False");
|
2008-02-28 01:57:57 +00:00
|
|
|
asprintf(&pfx, "%s\t", pfx2);
|
|
|
|
free(pfx2);
|
|
|
|
|
2009-07-13 14:49:20 +00:00
|
|
|
for (i=0; i < src->apps->size; i++) {
|
|
|
|
if (NULL == (app = (orte_app_context_t*)opal_pointer_array_get_item(src->apps, i))) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
opal_dss.print(&tmp2, pfx, app, ORTE_APP_CONTEXT);
|
2008-02-28 01:57:57 +00:00
|
|
|
asprintf(&tmp3, "%s\n%s", tmp, tmp2);
|
|
|
|
free(tmp);
|
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (NULL != src->map) {
|
|
|
|
if (ORTE_SUCCESS != (rc = opal_dss.print(&tmp2, pfx, src->map, ORTE_JOB_MAP))) {
|
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
|
|
|
free(tmp);
|
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
|
|
|
} else {
|
|
|
|
asprintf(&tmp2, "%s\n%sNo Map", tmp, pfx);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
|
2010-06-07 14:21:08 +00:00
|
|
|
asprintf(&tmp2, "%s\n%sNum procs: %ld", tmp, pfx, (long)src->num_procs);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
|
|
|
for (i=0; i < src->procs->size; i++) {
|
2009-07-13 14:49:20 +00:00
|
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(src->procs, i))) {
|
|
|
|
continue;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
2009-07-13 14:49:20 +00:00
|
|
|
if (ORTE_SUCCESS != (rc = opal_dss.print(&tmp2, pfx, proc, ORTE_PROC))) {
|
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
|
|
|
free(tmp);
|
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tNum launched: %ld\tNum reported: %ld\tNum terminated: %ld",
|
|
|
|
tmp, pfx, (long)src->num_launched, (long)src->num_reported,
|
|
|
|
(long)src->num_terminated);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
|
|
|
/* set the return */
|
|
|
|
*output = tmp;
|
|
|
|
free(pfx);
|
|
|
|
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* NODE
|
|
|
|
*/
|
|
|
|
int orte_dt_print_node(char **output, char *prefix, orte_node_t *src, opal_data_type_t type)
|
|
|
|
{
|
|
|
|
char *tmp, *tmp2, *tmp3, *pfx2, *pfx;
|
|
|
|
int32_t i;
|
|
|
|
int rc;
|
2009-07-13 14:49:20 +00:00
|
|
|
orte_proc_t *proc;
|
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* protect against NULL prefix */
|
|
|
|
if (NULL == prefix) {
|
|
|
|
asprintf(&pfx2, " ");
|
|
|
|
} else {
|
|
|
|
asprintf(&pfx2, "%s", prefix);
|
|
|
|
}
|
2008-08-31 18:06:55 +00:00
|
|
|
|
2008-06-04 20:53:12 +00:00
|
|
|
if (orte_xml_output) {
|
|
|
|
/* need to create the output in XML format */
|
2009-01-15 18:11:50 +00:00
|
|
|
asprintf(&tmp, "%s<host name=\"%s\" slots=\"%d\" max_slots=\"%d\">\n", pfx2,
|
2008-06-04 20:53:12 +00:00
|
|
|
(NULL == src->name) ? "UNKNOWN" : src->name,
|
|
|
|
(int)src->slots, (int)src->slots_max);
|
2009-01-15 18:11:50 +00:00
|
|
|
/* does this node have any aliases? */
|
|
|
|
if (NULL != src->alias) {
|
|
|
|
for (i=0; NULL != src->alias[i]; i++) {
|
|
|
|
asprintf(&tmp2, "%s%s\t<noderesolve resolved=\"%s\"/>\n", tmp, pfx2, src->alias[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*output = tmp;
|
2008-06-04 20:53:12 +00:00
|
|
|
free(pfx2);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
2008-02-28 01:57:57 +00:00
|
|
|
|
2008-09-23 15:46:34 +00:00
|
|
|
if (!orte_devel_level_output) {
|
|
|
|
/* just provide a simple output for users */
|
2008-09-28 16:44:09 +00:00
|
|
|
if (0 == src->num_procs) {
|
|
|
|
/* no procs mapped yet, so just show allocation */
|
2009-07-14 14:34:11 +00:00
|
|
|
asprintf(&tmp, "\n%sData for node: %s\tNum slots: %ld\tMax slots: %ld",
|
2008-09-28 16:44:09 +00:00
|
|
|
pfx2, (NULL == src->name) ? "UNKNOWN" : src->name,
|
|
|
|
(long)src->slots, (long)src->slots_max);
|
2009-01-15 18:11:50 +00:00
|
|
|
/* does this node have any aliases? */
|
|
|
|
if (NULL != src->alias) {
|
|
|
|
for (i=0; NULL != src->alias[i]; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tresolved from %s", tmp, pfx2, src->alias[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
}
|
2008-09-28 16:44:09 +00:00
|
|
|
free(pfx2);
|
|
|
|
*output = tmp;
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
2009-07-14 14:34:11 +00:00
|
|
|
asprintf(&tmp, "\n%sData for node: %s\tNum procs: %ld",
|
2008-09-23 15:46:34 +00:00
|
|
|
pfx2, (NULL == src->name) ? "UNKNOWN" : src->name,
|
|
|
|
(long)src->num_procs);
|
2009-01-15 18:11:50 +00:00
|
|
|
/* does this node have any aliases? */
|
|
|
|
if (NULL != src->alias) {
|
|
|
|
for (i=0; NULL != src->alias[i]; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tresolved from %s", tmp, pfx2, src->alias[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
}
|
2008-09-23 15:46:34 +00:00
|
|
|
goto PRINT_PROCS;
|
|
|
|
}
|
|
|
|
|
2009-07-14 14:34:11 +00:00
|
|
|
asprintf(&tmp, "\n%sData for node: %s\t%s\tLaunch id: %ld\tState: %0x",
|
2008-09-10 01:59:49 +00:00
|
|
|
pfx2, (NULL == src->name) ? "UNKNOWN" : src->name,
|
2008-02-28 01:57:57 +00:00
|
|
|
pfx2, (long)src->launch_id,
|
2009-07-13 20:03:41 +00:00
|
|
|
src->state);
|
2009-01-15 18:11:50 +00:00
|
|
|
/* does this node have any aliases? */
|
|
|
|
if (NULL != src->alias) {
|
|
|
|
for (i=0; NULL != src->alias[i]; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tresolved from %s", tmp, pfx2, src->alias[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
}
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
if (NULL == src->daemon) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tDaemon: %s\tDaemon launched: %s", tmp, pfx2,
|
|
|
|
"Not defined", src->daemon_launched ? "True" : "False");
|
|
|
|
} else {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tDaemon: %s\tDaemon launched: %s", tmp, pfx2,
|
2009-03-05 21:50:47 +00:00
|
|
|
ORTE_NAME_PRINT(&(src->daemon->name)), src->daemon_launched ? "True" : "False");
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
2010-12-01 12:51:39 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tNum slots: %ld\tSlots in use: %ld\tOversubscribed: %s", tmp, pfx2,
|
|
|
|
(long)src->slots, (long)src->slots_inuse,
|
|
|
|
(src->oversubscribed) ? "TRUE" : "FALSE");
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tNum slots allocated: %ld\tMax slots: %ld", tmp, pfx2,
|
2012-09-05 01:30:39 +00:00
|
|
|
(long)src->slots, (long)src->slots_max);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
|
|
|
asprintf(&tmp2, "%s\n%s\tUsername on node: %s", tmp, pfx2,
|
|
|
|
(NULL == src->username) ? "NULL" : src->username);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
2011-09-11 19:02:24 +00:00
|
|
|
#if OPAL_HAVE_HWLOC
|
2011-10-29 15:12:45 +00:00
|
|
|
if (orte_display_topo_with_map && NULL != src->topology) {
|
2011-09-11 19:02:24 +00:00
|
|
|
char *pfx3;
|
2011-10-30 13:20:16 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tDetected Resources:\n", tmp, pfx2);
|
2011-09-11 19:02:24 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
2010-01-25 22:25:13 +00:00
|
|
|
|
2011-09-11 19:02:24 +00:00
|
|
|
tmp2 = NULL;
|
2011-10-30 13:20:16 +00:00
|
|
|
asprintf(&pfx3, "%s\t\t", pfx2);
|
2011-09-11 19:02:24 +00:00
|
|
|
opal_dss.print(&tmp2, pfx3, src->topology, OPAL_HWLOC_TOPO);
|
|
|
|
free(pfx3);
|
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
2010-01-25 22:25:13 +00:00
|
|
|
free(tmp);
|
2011-09-11 19:02:24 +00:00
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
2009-11-30 23:11:25 +00:00
|
|
|
}
|
2011-09-11 19:02:24 +00:00
|
|
|
#endif
|
|
|
|
|
2008-04-30 19:49:53 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tNum procs: %ld\tNext node_rank: %ld", tmp, pfx2,
|
|
|
|
(long)src->num_procs, (long)src->next_node_rank);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
2008-09-23 15:46:34 +00:00
|
|
|
PRINT_PROCS:
|
2008-02-28 01:57:57 +00:00
|
|
|
asprintf(&pfx, "%s\t", pfx2);
|
|
|
|
free(pfx2);
|
|
|
|
|
|
|
|
for (i=0; i < src->procs->size; i++) {
|
2009-07-13 14:49:20 +00:00
|
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(src->procs, i))) {
|
|
|
|
continue;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
2009-07-13 14:49:20 +00:00
|
|
|
if (ORTE_SUCCESS != (rc = opal_dss.print(&tmp2, pfx, proc, ORTE_PROC))) {
|
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
|
|
|
free(tmp);
|
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
free(pfx);
|
2009-07-13 14:49:20 +00:00
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
/* set the return */
|
|
|
|
*output = tmp;
|
|
|
|
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* PROC
|
|
|
|
*/
|
|
|
|
int orte_dt_print_proc(char **output, char *prefix, orte_proc_t *src, opal_data_type_t type)
|
|
|
|
{
|
|
|
|
char *tmp, *tmp2, *pfx2;
|
2011-10-29 15:12:45 +00:00
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* protect against NULL prefix */
|
|
|
|
if (NULL == prefix) {
|
|
|
|
asprintf(&pfx2, " ");
|
|
|
|
} else {
|
|
|
|
asprintf(&pfx2, "%s", prefix);
|
|
|
|
}
|
|
|
|
|
2008-06-04 20:53:12 +00:00
|
|
|
if (orte_xml_output) {
|
|
|
|
/* need to create the output in XML format */
|
2010-04-26 22:15:57 +00:00
|
|
|
if (0 == src->pid) {
|
2011-08-26 22:16:14 +00:00
|
|
|
asprintf(output, "%s<process rank=\"%s\" status=\"%s\"/>\n", pfx2,
|
|
|
|
ORTE_VPID_PRINT(src->name.vpid), orte_proc_state_to_str(src->state));
|
2008-06-04 20:53:12 +00:00
|
|
|
} else {
|
2011-08-26 22:16:14 +00:00
|
|
|
asprintf(output, "%s<process rank=\"%s\" pid=\"%d\" status=\"%s\"/>\n", pfx2,
|
|
|
|
ORTE_VPID_PRINT(src->name.vpid), (int)src->pid, orte_proc_state_to_str(src->state));
|
2008-06-04 20:53:12 +00:00
|
|
|
}
|
|
|
|
free(pfx2);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
2008-09-23 15:46:34 +00:00
|
|
|
|
|
|
|
if (!orte_devel_level_output) {
|
|
|
|
/* just print a very simple output for users */
|
2011-12-05 22:01:08 +00:00
|
|
|
asprintf(&tmp, "\n%sProcess OMPI jobid: %s App: %ld Process rank: %s", pfx2,
|
|
|
|
ORTE_JOBID_PRINT(src->name.jobid), (long)src->app_idx,
|
2011-08-26 22:16:14 +00:00
|
|
|
ORTE_VPID_PRINT(src->name.vpid));
|
|
|
|
|
2008-09-23 15:46:34 +00:00
|
|
|
/* set the return */
|
|
|
|
*output = tmp;
|
|
|
|
free(pfx2);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
2008-06-04 20:53:12 +00:00
|
|
|
|
2009-03-05 21:50:47 +00:00
|
|
|
asprintf(&tmp, "\n%sData for proc: %s", pfx2, ORTE_NAME_PRINT(&src->name));
|
2008-02-28 01:57:57 +00:00
|
|
|
|
2011-06-16 20:31:30 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tPid: %ld\tLocal rank: %lu\tNode rank: %lu\tApp rank: %d", tmp, pfx2,
|
|
|
|
(long)src->pid, (unsigned long)src->local_rank, (unsigned long)src->node_rank, src->app_rank);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
2011-10-29 15:12:45 +00:00
|
|
|
#if OPAL_HAVE_HWLOC
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
{
|
|
|
|
char *locale=NULL;
|
|
|
|
|
|
|
|
if (NULL != src->locale) {
|
|
|
|
hwloc_bitmap_list_asprintf(&locale, src->locale->cpuset);
|
|
|
|
}
|
|
|
|
asprintf(&tmp2, "%s\n%s\tState: %s\tRestarts: %d\tApp_context: %ld\tLocale: %s\tBinding: %s[%u]", tmp, pfx2,
|
|
|
|
orte_proc_state_to_str(src->state), src->restarts, (long)src->app_idx,
|
|
|
|
(NULL == locale) ? "UNKNOWN" : locale,
|
|
|
|
(NULL == src->cpu_bitmap) ? "NULL" : src->cpu_bitmap, src->bind_idx);
|
|
|
|
if (NULL != locale) {
|
|
|
|
free(locale);
|
|
|
|
}
|
2011-10-29 15:12:45 +00:00
|
|
|
}
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
#else
|
|
|
|
asprintf(&tmp2, "%s\n%s\tState: %s\tRestarts: %d\tApp_context: %ld", tmp, pfx2,
|
|
|
|
orte_proc_state_to_str(src->state), src->restarts, (long)src->app_idx);
|
2011-10-29 15:12:45 +00:00
|
|
|
#endif
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
|
|
|
|
/* set the return */
|
|
|
|
*output = tmp2;
|
|
|
|
|
|
|
|
free(pfx2);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* APP CONTEXT
|
|
|
|
*/
|
|
|
|
int orte_dt_print_app_context(char **output, char *prefix, orte_app_context_t *src, opal_data_type_t type)
|
|
|
|
{
|
2008-03-05 22:12:27 +00:00
|
|
|
char *tmp, *tmp2, *pfx2;
|
|
|
|
int i, count;
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* protect against NULL prefix */
|
|
|
|
if (NULL == prefix) {
|
|
|
|
asprintf(&pfx2, " ");
|
|
|
|
} else {
|
|
|
|
asprintf(&pfx2, "%s", prefix);
|
|
|
|
}
|
|
|
|
|
2012-08-12 01:28:23 +00:00
|
|
|
asprintf(&tmp, "\n%sData for app_context: index %lu\tapp: %s\n%s\tNum procs: %lu\tFirstRank: %s\tRecovery: %s\tMax Restarts: %d",
|
2011-04-28 00:20:49 +00:00
|
|
|
pfx2, (unsigned long)src->idx,
|
2011-03-04 18:46:43 +00:00
|
|
|
(NULL == src->app) ? "NULL" : src->app,
|
2011-04-18 14:46:42 +00:00
|
|
|
pfx2, (unsigned long)src->num_procs,
|
2012-08-12 01:28:23 +00:00
|
|
|
ORTE_VPID_PRINT(src->first_rank),
|
2011-04-18 14:46:42 +00:00
|
|
|
(src->recovery_defined) ? "DEFINED" : "DEFAULT",
|
|
|
|
src->max_restarts);
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
count = opal_argv_count(src->argv);
|
|
|
|
for (i=0; i < count; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tArgv[%d]: %s", tmp, pfx2, i, src->argv[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
|
|
|
|
count = opal_argv_count(src->env);
|
|
|
|
for (i=0; i < count; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tEnv[%lu]: %s", tmp, pfx2, (unsigned long)i, src->env[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
|
2012-08-23 21:28:05 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tWorking dir: %s (user: %d session-dir: %d)\n%s\tPrefix: %s\n%s\tHostfile: %s\tAdd-Hostfile: %s", tmp,
|
|
|
|
pfx2, (NULL == src->cwd) ? "NULL" : src->cwd, (int) src->user_specified_cwd, (int) src->set_cwd_to_session_dir,
|
2009-06-10 19:01:08 +00:00
|
|
|
pfx2, (NULL == src->prefix_dir) ? "NULL" : src->prefix_dir,
|
2008-02-28 01:57:57 +00:00
|
|
|
pfx2, (NULL == src->hostfile) ? "NULL" : src->hostfile,
|
2008-03-05 22:12:27 +00:00
|
|
|
(NULL == src->add_hostfile) ? "NULL" : src->add_hostfile);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
|
2009-07-14 14:34:11 +00:00
|
|
|
count = opal_argv_count(src->add_host);
|
|
|
|
for (i=0; i < count; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tAdd_host[%lu]: %s", tmp, pfx2, (unsigned long)i, src->add_host[i]);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
|
|
|
|
2008-03-05 22:12:27 +00:00
|
|
|
count = opal_argv_count(src->dash_host);
|
|
|
|
for (i=0; i < count; i++) {
|
|
|
|
asprintf(&tmp2, "%s\n%s\tDash_host[%lu]: %s", tmp, pfx2, (unsigned long)i, src->dash_host[i]);
|
2008-02-28 01:57:57 +00:00
|
|
|
free(tmp);
|
2008-03-05 22:12:27 +00:00
|
|
|
tmp = tmp2;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
|
2012-08-24 02:28:29 +00:00
|
|
|
asprintf(&tmp2, "%s\n%s\tPreload binary: %s\tPreload files: %s\tUsed on node: %s", tmp,
|
|
|
|
pfx2, (src->preload_binary) ? "TRUE" : "FALSE",
|
|
|
|
(NULL == src->preload_files) ? "NULL" : src->preload_files,
|
|
|
|
(src->used_on_node) ? "TRUE" : "FALSE");
|
2009-06-10 19:01:08 +00:00
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
A number of C/R enhancements per RFC below:
http://www.open-mpi.org/community/lists/devel/2010/07/8240.php
Documentation:
http://osl.iu.edu/research/ft/
Major Changes:
--------------
* Added C/R-enabled Debugging support.
Enabled with the --enable-crdebug flag. See the following website for more information:
http://osl.iu.edu/research/ft/crdebug/
* Added Stable Storage (SStore) framework for checkpoint storage
* 'central' component does a direct to central storage save
* 'stage' component stages checkpoints to central storage while the application continues execution.
* 'stage' supports offline compression of checkpoints before moving (sstore_stage_compress)
* 'stage' supports local caching of checkpoints to improve automatic recovery (sstore_stage_caching)
* Added Compression (compress) framework to support
* Add two new ErrMgr recovery policies
* {{{crmig}}} C/R Process Migration
* {{{autor}}} C/R Automatic Recovery
* Added the {{{ompi-migrate}}} command line tool to support the {{{crmig}}} ErrMgr component
* Added CR MPI Ext functions (enable them with {{{--enable-mpi-ext=cr}}} configure option)
* {{{OMPI_CR_Checkpoint}}} (Fixes trac:2342)
* {{{OMPI_CR_Restart}}}
* {{{OMPI_CR_Migrate}}} (may need some more work for mapping rules)
* {{{OMPI_CR_INC_register_callback}}} (Fixes trac:2192)
* {{{OMPI_CR_Quiesce_start}}}
* {{{OMPI_CR_Quiesce_checkpoint}}}
* {{{OMPI_CR_Quiesce_end}}}
* {{{OMPI_CR_self_register_checkpoint_callback}}}
* {{{OMPI_CR_self_register_restart_callback}}}
* {{{OMPI_CR_self_register_continue_callback}}}
* The ErrMgr predicted_fault() interface has been changed to take an opal_list_t of ErrMgr defined types. This will allow us to better support a wider range of fault prediction services in the future.
* Add a progress meter to:
* FileM rsh (filem_rsh_process_meter)
* SnapC full (snapc_full_progress_meter)
* SStore stage (sstore_stage_progress_meter)
* Added 2 new command line options to ompi-restart
* --showme : Display the full command line that would have been exec'ed.
* --mpirun_opts : Command line options to pass directly to mpirun. (Fixes trac:2413)
* Deprecated some MCA params:
* crs_base_snapshot_dir deprecated, use sstore_stage_local_snapshot_dir
* snapc_base_global_snapshot_dir deprecated, use sstore_base_global_snapshot_dir
* snapc_base_global_shared deprecated, use sstore_stage_global_is_shared
* snapc_base_store_in_place deprecated, replaced with different components of SStore
* snapc_base_global_snapshot_ref deprecated, use sstore_base_global_snapshot_ref
* snapc_base_establish_global_snapshot_dir deprecated, never well supported
* snapc_full_skip_filem deprecated, use sstore_stage_skip_filem
Minor Changes:
--------------
* Fixes trac:1924 : {{{ompi-restart}}} now recognizes path prefixed checkpoint handles and does the right thing.
* Fixes trac:2097 : {{{ompi-info}}} should now report all available CRS components
* Fixes trac:2161 : Manual checkpoint movement. A user can 'mv' a checkpoint directory from the original location to another and still restart from it.
* Fixes trac:2208 : Honor various TMPDIR varaibles instead of forcing {{{/tmp}}}
* Move {{{ompi_cr_continue_like_restart}}} to {{{orte_cr_continue_like_restart}}} to be more flexible in where this should be set.
* opal_crs_base_metadata_write* functions have been moved to SStore to support a wider range of metadata handling functionality.
* Cleanup the CRS framework and components to work with the SStore framework.
* Cleanup the SnapC framework and components to work with the SStore framework (cleans up these code paths considerably).
* Add 'quiesce' hook to CRCP for a future enhancement.
* We now require a BLCR version that supports {{{cr_request_file()}}} or {{{cr_request_checkpoint()}}} in order to make the code more maintainable. Note that {{{cr_request_file}}} has been deprecated since 0.7.0, so we prefer to use {{{cr_request_checkpoint()}}}.
* Add optional application level INC callbacks (registered through the CR MPI Ext interface).
* Increase the {{{opal_cr_thread_sleep_wait}}} parameter to 1000 microseconds to make the C/R thread less aggressive.
* {{{opal-restart}}} now looks for cache directories before falling back on stable storage when asked.
* {{{opal-restart}}} also support local decompression before restarting
* {{{orte-checkpoint}}} now uses the SStore framework to work with the metadata
* {{{orte-restart}}} now uses the SStore framework to work with the metadata
* Remove the {{{orte-restart}}} preload option. This was removed since the user only needs to select the 'stage' component in order to support this functionality.
* Since the '-am' parameter is saved in the metadata, {{{ompi-restart}}} no longer hard codes {{{-am ft-enable-cr}}}.
* Fix {{{hnp}}} ErrMgr so that if a previous component in the stack has 'fixed' the problem, then it should be skipped.
* Make sure to decrement the number of 'num_local_procs' in the orted when one goes away.
* odls now checks the SStore framework to see if it needs to load any checkpoint files before launching (to support 'stage'). This separates the SStore logic from the --preload-[binary|files] options.
* Add unique IDs to the named pipes established between the orted and the app in SnapC. This is to better support migration and automatic recovery activities.
* Improve the checks for 'already checkpointing' error path.
* A a recovery output timer, to show how long it takes to restart a job
* Do a better job of cleaning up the old session directory on restart.
* Add a local module to the autor and crmig ErrMgr components. These small modules prevent the 'orted' component from attempting a local recovery (Which does not work for MPI apps at the moment)
* Add a fix for bounding the checkpointable region between MPI_Init and MPI_Finalize.
This commit was SVN r23587.
The following Trac tickets were found above:
Ticket 1924 --> https://svn.open-mpi.org/trac/ompi/ticket/1924
Ticket 2097 --> https://svn.open-mpi.org/trac/ompi/ticket/2097
Ticket 2161 --> https://svn.open-mpi.org/trac/ompi/ticket/2161
Ticket 2192 --> https://svn.open-mpi.org/trac/ompi/ticket/2192
Ticket 2208 --> https://svn.open-mpi.org/trac/ompi/ticket/2208
Ticket 2342 --> https://svn.open-mpi.org/trac/ompi/ticket/2342
Ticket 2413 --> https://svn.open-mpi.org/trac/ompi/ticket/2413
2010-08-10 20:51:11 +00:00
|
|
|
|
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
|
|
|
asprintf(&tmp2, "%s\n%s\tSStore Load: %s", tmp,
|
|
|
|
pfx2, src->sstore_load);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
#endif
|
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
/* set the return */
|
|
|
|
*output = tmp;
|
|
|
|
|
|
|
|
free(pfx2);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* JOB_MAP
|
|
|
|
*/
|
|
|
|
int orte_dt_print_map(char **output, char *prefix, orte_job_map_t *src, opal_data_type_t type)
|
|
|
|
{
|
2008-09-23 15:46:34 +00:00
|
|
|
char *tmp=NULL, *tmp2, *tmp3, *pfx, *pfx2;
|
2008-06-04 20:53:12 +00:00
|
|
|
int32_t i, j;
|
2008-02-28 01:57:57 +00:00
|
|
|
int rc;
|
2009-07-13 14:49:20 +00:00
|
|
|
orte_node_t *node;
|
|
|
|
orte_proc_t *proc;
|
2008-02-28 01:57:57 +00:00
|
|
|
|
|
|
|
/* set default result */
|
|
|
|
*output = NULL;
|
|
|
|
|
|
|
|
/* protect against NULL prefix */
|
|
|
|
if (NULL == prefix) {
|
|
|
|
asprintf(&pfx2, " ");
|
|
|
|
} else {
|
|
|
|
asprintf(&pfx2, "%s", prefix);
|
|
|
|
}
|
|
|
|
|
2008-06-04 20:53:12 +00:00
|
|
|
if (orte_xml_output) {
|
|
|
|
/* need to create the output in XML format */
|
|
|
|
asprintf(&tmp, "<map>\n");
|
|
|
|
/* loop through nodes */
|
|
|
|
for (i=0; i < src->nodes->size; i++) {
|
2009-07-13 14:49:20 +00:00
|
|
|
if (NULL == (node = (orte_node_t*)opal_pointer_array_get_item(src->nodes, i))) {
|
|
|
|
continue;
|
2008-06-04 20:53:12 +00:00
|
|
|
}
|
2009-07-13 14:49:20 +00:00
|
|
|
orte_dt_print_node(&tmp2, "\t", node, ORTE_NODE);
|
2008-06-04 20:53:12 +00:00
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
|
|
|
free(tmp2);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp3;
|
|
|
|
/* for each node, loop through procs and print their rank */
|
2009-07-13 14:49:20 +00:00
|
|
|
for (j=0; j < node->procs->size; j++) {
|
|
|
|
if (NULL == (proc = (orte_proc_t*)opal_pointer_array_get_item(node->procs, j))) {
|
|
|
|
continue;
|
2008-06-04 20:53:12 +00:00
|
|
|
}
|
2009-07-13 14:49:20 +00:00
|
|
|
orte_dt_print_proc(&tmp2, "\t\t", proc, ORTE_PROC);
|
2008-06-04 20:53:12 +00:00
|
|
|
asprintf(&tmp3, "%s%s", tmp, tmp2);
|
|
|
|
free(tmp2);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp3;
|
|
|
|
}
|
|
|
|
asprintf(&tmp3, "%s\t</host>\n", tmp);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp3;
|
|
|
|
}
|
|
|
|
asprintf(&tmp2, "%s</map>\n", tmp);
|
|
|
|
free(tmp);
|
|
|
|
free(pfx2);
|
|
|
|
*output = tmp2;
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
|
|
|
|
}
|
2008-08-31 18:06:55 +00:00
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
asprintf(&pfx, "%s\t", pfx2);
|
|
|
|
|
2008-09-23 15:46:34 +00:00
|
|
|
if (orte_devel_level_output) {
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
#if OPAL_HAVE_HWLOC
|
|
|
|
asprintf(&tmp, "\n%sMapper requested: %s Last mapper: %s Mapping policy: %s Ranking policy: %s Binding policy: %s[%s] Cpu set: %s PPR: %s",
|
2011-03-12 05:30:09 +00:00
|
|
|
pfx2, (NULL == src->req_mapper) ? "NULL" : src->req_mapper,
|
|
|
|
(NULL == src->last_mapper) ? "NULL" : src->last_mapper,
|
At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement
The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.
In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:
1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.
2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.
3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.
As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.
This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
|
|
|
orte_rmaps_base_print_mapping(src->mapping),
|
|
|
|
orte_rmaps_base_print_ranking(src->ranking),
|
|
|
|
opal_hwloc_base_print_binding(src->binding),
|
|
|
|
opal_hwloc_base_print_level(src->bind_level),
|
|
|
|
(NULL == opal_hwloc_base_cpu_set) ? "NULL" : opal_hwloc_base_cpu_set,
|
|
|
|
(NULL == src->ppr) ? "NULL" : src->ppr);
|
|
|
|
#else
|
|
|
|
asprintf(&tmp, "\n%sMapper requested: %s Last mapper: %s Mapping policy: %s Ranking policy: %s PPR: %s",
|
|
|
|
pfx2, (NULL == src->req_mapper) ? "NULL" : src->req_mapper,
|
|
|
|
(NULL == src->last_mapper) ? "NULL" : src->last_mapper,
|
|
|
|
orte_rmaps_base_print_mapping(src->mapping),
|
|
|
|
orte_rmaps_base_print_ranking(src->ranking),
|
|
|
|
(NULL == src->ppr) ? "NULL" : src->ppr);
|
|
|
|
#endif
|
|
|
|
|
2008-09-23 15:46:34 +00:00
|
|
|
if (ORTE_VPID_INVALID == src->daemon_vpid_start) {
|
|
|
|
asprintf(&tmp2, "%s\n%sNum new daemons: %ld\tNew daemon starting vpid INVALID\n%sNum nodes: %ld",
|
|
|
|
tmp, pfx, (long)src->num_new_daemons, pfx, (long)src->num_nodes);
|
|
|
|
} else {
|
|
|
|
asprintf(&tmp2, "%s\n%sNum new daemons: %ld\tNew daemon starting vpid %ld\n%sNum nodes: %ld",
|
|
|
|
tmp, pfx, (long)src->num_new_daemons, (long)src->daemon_vpid_start,
|
|
|
|
pfx, (long)src->num_nodes);
|
|
|
|
}
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
2008-09-28 16:44:09 +00:00
|
|
|
} else {
|
|
|
|
/* this is being printed for a user, so let's make it easier to see */
|
|
|
|
asprintf(&tmp, "\n%s======================== JOB MAP ========================", pfx2);
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
for (i=0; i < src->nodes->size; i++) {
|
2009-07-13 14:49:20 +00:00
|
|
|
if (NULL == (node = (orte_node_t*)opal_pointer_array_get_item(src->nodes, i))) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (ORTE_SUCCESS != (rc = opal_dss.print(&tmp2, pfx2, node, ORTE_NODE))) {
|
|
|
|
ORTE_ERROR_LOG(rc);
|
|
|
|
free(pfx);
|
2008-10-02 20:08:27 +00:00
|
|
|
free(tmp);
|
2009-07-13 14:49:20 +00:00
|
|
|
return rc;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
2009-07-13 14:49:20 +00:00
|
|
|
asprintf(&tmp3, "%s\n%s", tmp, tmp2);
|
|
|
|
free(tmp);
|
|
|
|
free(tmp2);
|
|
|
|
tmp = tmp3;
|
2008-02-28 01:57:57 +00:00
|
|
|
}
|
|
|
|
|
2008-09-28 16:44:09 +00:00
|
|
|
if (!orte_devel_level_output) {
|
|
|
|
/* this is being printed for a user, so let's make it easier to see */
|
|
|
|
asprintf(&tmp2, "%s\n\n%s=============================================================\n", tmp, pfx2);
|
|
|
|
free(tmp);
|
|
|
|
tmp = tmp2;
|
|
|
|
}
|
2008-10-02 20:08:27 +00:00
|
|
|
free(pfx2);
|
2008-09-28 16:44:09 +00:00
|
|
|
|
2008-02-28 01:57:57 +00:00
|
|
|
/* set the return */
|
|
|
|
*output = tmp;
|
|
|
|
|
|
|
|
free(pfx);
|
|
|
|
return ORTE_SUCCESS;
|
|
|
|
}
|