2009-06-30 21:41:28 +00:00
|
|
|
/*
|
2010-03-12 23:57:50 +00:00
|
|
|
* Copyright (c) 2004-2010 The Trustees of Indiana University and Indiana
|
2009-06-30 21:41:28 +00:00
|
|
|
* University Research and Technology
|
|
|
|
* Corporation. All rights reserved.
|
|
|
|
* Copyright (c) 2004-2005 The University of Tennessee and The University
|
|
|
|
* 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.
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
* Copyright (c) 2006-2010 Cisco Systems, Inc. All rights reserved.
|
2011-06-21 15:41:57 +00:00
|
|
|
* Copyright (c) 2010 Los Alamos National Security, LLC.
|
|
|
|
* All rights reserved.
|
2011-09-07 14:23:28 +00:00
|
|
|
* Copyright (c) 2011 University of Houston. All rights reserved.
|
2009-06-30 21:41:28 +00:00
|
|
|
* $COPYRIGHT$
|
|
|
|
*
|
|
|
|
* Additional copyrights may follow
|
|
|
|
*
|
|
|
|
* $HEADER$
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "ompi_config.h"
|
|
|
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
|
|
|
|
#include "ompi/runtime/params.h"
|
|
|
|
#include "orte/runtime/runtime.h"
|
|
|
|
|
|
|
|
#include "opal/util/argv.h"
|
|
|
|
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
#include "opal/mca/event/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "opal/util/output.h"
|
|
|
|
#include "opal/mca/base/base.h"
|
|
|
|
#include "opal/mca/backtrace/backtrace.h"
|
|
|
|
#include "opal/mca/backtrace/base/base.h"
|
|
|
|
#include "opal/mca/paffinity/paffinity.h"
|
|
|
|
#include "opal/mca/paffinity/base/base.h"
|
|
|
|
#include "opal/mca/carto/carto.h"
|
|
|
|
#include "opal/mca/carto/base/base.h"
|
2011-06-21 15:41:57 +00:00
|
|
|
#include "opal/mca/shmem/shmem.h"
|
|
|
|
#include "opal/mca/shmem/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "opal/mca/maffinity/maffinity.h"
|
|
|
|
#include "opal/mca/maffinity/base/base.h"
|
|
|
|
#include "opal/mca/memory/memory.h"
|
|
|
|
#include "opal/mca/memory/base/base.h"
|
|
|
|
#include "opal/mca/memchecker/memchecker.h"
|
|
|
|
#include "opal/mca/memchecker/base/base.h"
|
|
|
|
#include "opal/mca/timer/timer.h"
|
|
|
|
#include "opal/mca/timer/base/base.h"
|
|
|
|
#include "opal/mca/installdirs/installdirs.h"
|
|
|
|
#include "opal/mca/installdirs/base/base.h"
|
2011-09-11 19:02:24 +00:00
|
|
|
#include "opal/mca/hwloc/base/base.h"
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "opal/mca/crs/crs.h"
|
|
|
|
#include "opal/mca/crs/base/base.h"
|
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
|
|
|
#include "opal/mca/compress/compress.h"
|
|
|
|
#include "opal/mca/compress/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#endif
|
|
|
|
#include "opal/runtime/opal.h"
|
|
|
|
#include "opal/dss/dss.h"
|
2010-09-22 01:11:40 +00:00
|
|
|
#include "opal/mca/if/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
|
|
|
|
#include "ompi/mca/allocator/base/base.h"
|
|
|
|
#include "ompi/mca/coll/base/base.h"
|
|
|
|
#include "ompi/mca/io/io.h"
|
|
|
|
#include "ompi/mca/io/base/base.h"
|
|
|
|
#include "ompi/mca/mpool/base/base.h"
|
|
|
|
#include "ompi/mca/pml/pml.h"
|
|
|
|
#include "ompi/mca/pml/base/base.h"
|
|
|
|
#include "ompi/mca/bml/base/base.h"
|
|
|
|
#include "ompi/mca/rcache/rcache.h"
|
|
|
|
#include "ompi/mca/rcache/base/base.h"
|
|
|
|
#include "ompi/mca/btl/base/base.h"
|
|
|
|
#include "ompi/mca/mtl/mtl.h"
|
|
|
|
#include "ompi/mca/mtl/base/base.h"
|
|
|
|
#include "ompi/mca/topo/base/base.h"
|
|
|
|
#include "ompi/mca/osc/osc.h"
|
|
|
|
#include "ompi/mca/osc/base/base.h"
|
|
|
|
#include "ompi/mca/pubsub/base/base.h"
|
|
|
|
#include "ompi/mca/dpm/base/base.h"
|
|
|
|
#include "ompi/mca/op/base/base.h"
|
2010-10-22 03:21:10 +00:00
|
|
|
#include "ompi/mca/vprotocol/base/base.h"
|
2011-09-07 14:23:28 +00:00
|
|
|
#include "ompi/mca/fbtl/fbtl.h"
|
|
|
|
#include "ompi/mca/fbtl/base/base.h"
|
|
|
|
#include "ompi/mca/fs/fs.h"
|
|
|
|
#include "ompi/mca/fs/base/base.h"
|
|
|
|
#include "ompi/mca/fcache/fcache.h"
|
|
|
|
#include "ompi/mca/fcache/base/base.h"
|
|
|
|
#include "ompi/mca/fcoll/fcoll.h"
|
|
|
|
#include "ompi/mca/fcoll/base/base.h"
|
|
|
|
#include "ompi/mca/sharedfp/sharedfp.h"
|
|
|
|
#include "ompi/mca/sharedfp/base/base.h"
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "ompi/mca/crcp/crcp.h"
|
|
|
|
#include "ompi/mca/crcp/base/base.h"
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#include "orte/mca/errmgr/errmgr.h"
|
|
|
|
#include "orte/mca/errmgr/base/base.h"
|
|
|
|
#include "orte/mca/grpcomm/grpcomm.h"
|
|
|
|
#include "orte/mca/grpcomm/base/base.h"
|
2010-05-03 04:11:03 +00:00
|
|
|
#include "orte/mca/db/db.h"
|
|
|
|
#include "orte/mca/db/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "orte/mca/ess/ess.h"
|
|
|
|
#include "orte/mca/ess/base/base.h"
|
|
|
|
#include "orte/mca/notifier/notifier.h"
|
|
|
|
#include "orte/mca/notifier/base/base.h"
|
|
|
|
#include "orte/util/show_help.h"
|
|
|
|
#include "orte/util/proc_info.h"
|
|
|
|
#if !ORTE_DISABLE_FULL_SUPPORT
|
2010-08-25 22:52:33 +00:00
|
|
|
#include "orte/mca/debugger/debugger.h"
|
|
|
|
#include "orte/mca/debugger/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "orte/mca/iof/iof.h"
|
|
|
|
#include "orte/mca/iof/base/base.h"
|
|
|
|
#include "orte/mca/oob/oob.h"
|
|
|
|
#include "orte/mca/oob/base/base.h"
|
|
|
|
#include "orte/mca/odls/odls.h"
|
|
|
|
#include "orte/mca/odls/base/base.h"
|
|
|
|
#include "orte/mca/ras/ras.h"
|
|
|
|
#include "orte/mca/ras/base/ras_private.h"
|
|
|
|
#include "orte/mca/rmaps/rmaps.h"
|
|
|
|
#include "orte/mca/rmaps/base/base.h"
|
2009-09-29 13:14:39 +00:00
|
|
|
#include "orte/mca/rmcast/rmcast.h"
|
|
|
|
#include "orte/mca/rmcast/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "orte/mca/rml/rml.h"
|
|
|
|
#include "orte/mca/rml/base/base.h"
|
|
|
|
#include "orte/mca/routed/routed.h"
|
|
|
|
#include "orte/mca/routed/base/base.h"
|
|
|
|
#include "orte/mca/plm/plm.h"
|
|
|
|
#include "orte/mca/plm/base/base.h"
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "orte/mca/snapc/snapc.h"
|
|
|
|
#include "orte/mca/snapc/base/base.h"
|
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
|
|
|
#include "orte/mca/sstore/sstore.h"
|
|
|
|
#include "orte/mca/sstore/base/base.h"
|
2009-06-30 21:41:28 +00:00
|
|
|
#endif
|
2010-04-26 22:15:57 +00:00
|
|
|
#if ORTE_ENABLE_SENSORS
|
|
|
|
#include "orte/mca/sensor/sensor.h"
|
|
|
|
#include "orte/mca/sensor/base/base.h"
|
|
|
|
#endif
|
2009-06-30 21:41:28 +00:00
|
|
|
#include "orte/mca/filem/filem.h"
|
|
|
|
#include "orte/mca/filem/base/base.h"
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#include "ompi/tools/ompi_info/ompi_info.h"
|
|
|
|
/*
|
|
|
|
* Public variables
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void component_map_construct(ompi_info_component_map_t *map)
|
|
|
|
{
|
|
|
|
map->type = NULL;
|
|
|
|
}
|
|
|
|
static void component_map_destruct(ompi_info_component_map_t *map)
|
|
|
|
{
|
|
|
|
if (NULL != map->type) {
|
|
|
|
free(map->type);
|
|
|
|
}
|
|
|
|
/* the type close functions will release the
|
|
|
|
* list of components
|
|
|
|
*/
|
|
|
|
}
|
|
|
|
OBJ_CLASS_INSTANCE(ompi_info_component_map_t,
|
|
|
|
opal_list_item_t,
|
|
|
|
component_map_construct,
|
|
|
|
component_map_destruct);
|
|
|
|
|
|
|
|
opal_pointer_array_t component_map;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Private variables
|
|
|
|
*/
|
|
|
|
|
|
|
|
static bool opened_components = false;
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Open all MCA components so that they can register their MCA
|
|
|
|
* parameters. Take a shotgun approach here and indiscriminately open
|
|
|
|
* all components -- don't be selective. To this end, we need to clear
|
|
|
|
* out the environment of all OMPI_MCA_<type> variables to ensure
|
|
|
|
* that the open algorithms don't try to only open one component.
|
|
|
|
*/
|
|
|
|
void ompi_info_open_components(void)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
char *env, *str;
|
|
|
|
char *target, *save, *type;
|
|
|
|
char **env_save=NULL;
|
|
|
|
bool need_close_components = false;
|
|
|
|
ompi_info_component_map_t *map;
|
|
|
|
|
|
|
|
if (opened_components) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* init the map */
|
|
|
|
OBJ_CONSTRUCT(&component_map, opal_pointer_array_t);
|
|
|
|
opal_pointer_array_init(&component_map, 256, INT_MAX, 128);
|
|
|
|
|
|
|
|
/* Clear out the environment. Use strdup() to orphan the resulting
|
|
|
|
* strings because items are placed in the environment by reference,
|
|
|
|
* not by value.
|
|
|
|
*/
|
|
|
|
|
|
|
|
for (i = 0; i < mca_types.size; ++i) {
|
|
|
|
if (NULL == (type = (char*)opal_pointer_array_get_item(&mca_types, i))) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
asprintf(&env, "OMPI_MCA_%s", type);
|
|
|
|
if (NULL != (save = getenv(env))) {
|
|
|
|
/* save this param so it can later be restored */
|
|
|
|
asprintf(&str, "%s=%s", env, save);
|
|
|
|
opal_argv_append_nosize(&env_save, str);
|
|
|
|
free(str);
|
|
|
|
/* can't manipulate it directly, so make a copy first */
|
|
|
|
asprintf(&target, "%s=", env);
|
|
|
|
putenv(target);
|
|
|
|
free(target);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* some components require the event library be active, so activate it */
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
if (OPAL_SUCCESS != opal_event_base_open()) {
|
|
|
|
str = "opal_event_base_open";
|
2009-06-30 21:41:28 +00:00
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Open the DSS */
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != opal_dss_open()) {
|
|
|
|
str = "Unable to initialize the DSS";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Open up the MCA */
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != mca_base_open()) {
|
|
|
|
str = "mca_base_open failed";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Register the OPAL layer's MCA parameters */
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_register_params()) {
|
|
|
|
str = "opal_register_params failed";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Register the ORTE layer's MCA parameters */
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_register_params()) {
|
|
|
|
str = "orte_register_params failed";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Initialize the opal_output system */
|
|
|
|
if (!opal_output_init()) {
|
|
|
|
str = "opal_output_init failed";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Register the MPI layer's MCA parameters */
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != ompi_mpi_register_params()) {
|
|
|
|
str = "ompi_mpi_Register_params failed";
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Find / open all components */
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("base");
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
/* set default error message from here forward */
|
|
|
|
str = "A component framework failed to open properly.";
|
|
|
|
|
|
|
|
/* OPAL frameworks */
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_backtrace_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("backtrace");
|
|
|
|
map->components = &opal_backtrace_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_memory_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("memory");
|
|
|
|
map->components = &opal_memory_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2010-10-26 02:41:42 +00:00
|
|
|
/* the event framework is already open - just get its components */
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("event");
|
|
|
|
map->components = &opal_event_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (OPAL_SUCCESS != opal_memchecker_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("memchecker");
|
|
|
|
map->components = &opal_memchecker_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_paffinity_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("paffinity");
|
|
|
|
map->components = &opal_paffinity_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_carto_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("carto");
|
|
|
|
map->components = &opal_carto_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2011-06-21 15:41:57 +00:00
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_shmem_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("shmem");
|
|
|
|
map->components = &opal_shmem_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2009-06-30 21:41:28 +00:00
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_maffinity_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("maffinity");
|
|
|
|
map->components = &opal_maffinity_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2011-09-11 19:02:24 +00:00
|
|
|
#if OPAL_HAVE_HWLOC
|
|
|
|
if (OPAL_SUCCESS != opal_hwloc_base_open()) {
|
2010-03-29 20:47:29 +00:00
|
|
|
goto error;
|
|
|
|
}
|
2010-03-29 21:54:02 +00:00
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
2011-09-11 19:02:24 +00:00
|
|
|
map->type = strdup("hwloc");
|
2011-09-17 11:54:36 +00:00
|
|
|
map->components = &opal_hwloc_base_components;
|
2010-03-29 20:47:29 +00:00
|
|
|
opal_pointer_array_add(&component_map, map);
|
2011-09-11 19:02:24 +00:00
|
|
|
#endif
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (OPAL_SUCCESS != opal_timer_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("timer");
|
|
|
|
map->components = &opal_timer_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
if (OPAL_SUCCESS != opal_crs_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("crs");
|
|
|
|
map->components = &opal_crs_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
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_SUCCESS != opal_compress_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("compress");
|
|
|
|
map->components = &opal_compress_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2009-06-30 21:41:28 +00:00
|
|
|
#endif
|
2010-09-22 01:11:40 +00:00
|
|
|
|
|
|
|
if (OPAL_SUCCESS != opal_if_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("if");
|
|
|
|
map->components = &opal_if_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
/* OPAL's installdirs base open has already been called as part of
|
|
|
|
* opal_init_util() back in main().
|
|
|
|
*/
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("installdirs");
|
|
|
|
map->components = &opal_installdirs_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
/* ORTE frameworks
|
|
|
|
* Set orte_process_info.proc_type to HNP to force all frameworks to
|
|
|
|
* open components
|
|
|
|
*/
|
|
|
|
orte_process_info.proc_type = ORTE_PROC_HNP;
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_errmgr_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("errmgr");
|
|
|
|
map->components = &orte_errmgr_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_grpcomm_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("grpcomm");
|
2010-02-25 01:11:29 +00:00
|
|
|
map->components = &orte_grpcomm_base.components_available;
|
2009-06-30 21:41:28 +00:00
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2010-05-03 04:11:03 +00:00
|
|
|
if (ORTE_SUCCESS != orte_db_base_open()) {
|
2010-03-18 02:12:59 +00:00
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
2010-05-03 04:11:03 +00:00
|
|
|
map->type = strdup("db");
|
|
|
|
map->components = &orte_db_base_components_available;
|
2010-03-18 02:12:59 +00:00
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (ORTE_SUCCESS != orte_ess_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("ess");
|
|
|
|
map->components = &orte_ess_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_notifier_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("notifier");
|
2010-05-17 23:08:56 +00:00
|
|
|
map->components = &orte_notifier_base_components_available;
|
2009-06-30 21:41:28 +00:00
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
#if !ORTE_DISABLE_FULL_SUPPORT
|
2010-08-25 22:52:33 +00:00
|
|
|
if (ORTE_SUCCESS != orte_debugger_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("debugger");
|
|
|
|
map->components = &orte_debugger_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (ORTE_SUCCESS != mca_oob_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("oob");
|
|
|
|
map->components = &mca_oob_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_odls_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("odls");
|
|
|
|
map->components = &orte_odls_base.available_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_iof_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("iof");
|
|
|
|
map->components = &orte_iof_base.iof_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_ras_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("ras");
|
|
|
|
map->components = &orte_ras_base.ras_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_rmaps_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("rmaps");
|
|
|
|
map->components = &orte_rmaps_base.available_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-09-29 13:14:39 +00:00
|
|
|
if (ORTE_SUCCESS != orte_rmcast_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("rmcast");
|
2009-10-23 02:38:51 +00:00
|
|
|
map->components = &orte_rmcast_base.rmcast_opened;
|
2009-09-29 13:14:39 +00:00
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (ORTE_SUCCESS != orte_rml_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("rml");
|
|
|
|
map->components = &orte_rml_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_routed_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("routed");
|
|
|
|
map->components = &orte_routed_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (ORTE_SUCCESS != orte_plm_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("plm");
|
|
|
|
map->components = &orte_plm_base.available_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2009-12-02 16:03:06 +00:00
|
|
|
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
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 (ORTE_SUCCESS != orte_sstore_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("sstore");
|
|
|
|
map->components = &orte_sstore_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (ORTE_SUCCESS != orte_snapc_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("snapc");
|
|
|
|
map->components = &orte_snapc_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
#endif
|
|
|
|
|
2010-04-26 22:15:57 +00:00
|
|
|
#if ORTE_ENABLE_SENSORS
|
|
|
|
if (ORTE_SUCCESS != orte_sensor_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
2010-10-26 19:53:17 +00:00
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
2010-04-26 22:15:57 +00:00
|
|
|
map->type = strdup("sensor");
|
|
|
|
map->components = &mca_sensor_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
#endif
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
if (ORTE_SUCCESS != orte_filem_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("filem");
|
|
|
|
map->components = &orte_filem_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/* MPI frameworks */
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_allocator_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("allocator");
|
|
|
|
map->components = &mca_allocator_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_coll_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("coll");
|
|
|
|
map->components = &mca_coll_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_io_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("io");
|
|
|
|
map->components = &mca_io_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2011-09-07 14:23:28 +00:00
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_fcoll_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("fcoll");
|
|
|
|
map->components = &mca_fcoll_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_fs_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("fs");
|
|
|
|
map->components = &mca_fs_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_fbtl_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("fbtl");
|
|
|
|
map->components = &mca_fbtl_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_fcache_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("fcache");
|
|
|
|
map->components = &mca_fcache_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_sharedfp_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("sharedfp");
|
|
|
|
map->components = &mca_sharedfp_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
2009-06-30 21:41:28 +00:00
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_rcache_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("rcache");
|
|
|
|
map->components = &mca_rcache_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_mpool_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("mpool");
|
|
|
|
map->components = &mca_mpool_base_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_pml_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("pml");
|
|
|
|
map->components = &mca_pml_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
/* No need to call the bml_base_open() because the ob1 pml calls it.
|
|
|
|
* mca_bml_base_open();
|
|
|
|
*/
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("bml");
|
|
|
|
map->components = &mca_bml_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != ompi_osc_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("osc");
|
|
|
|
map->components = &ompi_osc_base_open_components;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_btl_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("btl");
|
|
|
|
map->components = &mca_btl_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != ompi_mtl_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("mtl");
|
|
|
|
map->components = &ompi_mtl_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != mca_topo_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("topo");
|
|
|
|
map->components = &mca_topo_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != ompi_pubsub_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("pubsub");
|
|
|
|
map->components = &ompi_pubsub_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
if (OMPI_SUCCESS != ompi_dpm_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("dpm");
|
|
|
|
map->components = &ompi_dpm_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
|
|
|
ompi_op_base_open();
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("op");
|
|
|
|
map->components = &ompi_op_base_components_opened;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2010-10-22 03:21:10 +00:00
|
|
|
if (OMPI_SUCCESS != mca_vprotocol_base_open(NULL)) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("vprotocol");
|
|
|
|
map->components = &mca_vprotocol_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
if (OMPI_SUCCESS != ompi_crcp_base_open()) {
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
map = OBJ_NEW(ompi_info_component_map_t);
|
|
|
|
map->type = strdup("crcp");
|
|
|
|
map->components = &ompi_crcp_base_components_available;
|
|
|
|
opal_pointer_array_add(&component_map, map);
|
|
|
|
#endif
|
|
|
|
|
2009-08-03 16:44:12 +00:00
|
|
|
/* flag that we need to close components */
|
|
|
|
need_close_components = true;
|
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
/* Restore the environment to what it was before we started so that
|
|
|
|
* if users setenv OMPI_MCA_<framework name> to some value, they'll
|
|
|
|
* see that value when it is shown via --param output.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (NULL != env_save) {
|
|
|
|
for (i = 0; i < opal_argv_count(env_save); ++i) {
|
|
|
|
putenv(env_save[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* All done */
|
|
|
|
|
|
|
|
opened_components = true;
|
|
|
|
return;
|
|
|
|
|
|
|
|
error:
|
|
|
|
fprintf(stderr, "%s\n", str);
|
|
|
|
fprintf(stderr, "ompi_info will likely not display all configuration information\n");
|
|
|
|
if (need_close_components) {
|
|
|
|
opened_components = true;
|
|
|
|
ompi_info_close_components();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
void ompi_info_close_components()
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
ompi_info_component_map_t *map;
|
|
|
|
|
|
|
|
if (opened_components) {
|
|
|
|
|
|
|
|
/* Note that the order of shutdown here doesn't matter because
|
|
|
|
* we aren't *using* any components -- none were selected, so
|
|
|
|
* there are no dependencies between the frameworks. We list
|
|
|
|
* them generally "in order", but it doesn't really matter.
|
|
|
|
|
|
|
|
* We also explicitly ignore the return values from the
|
|
|
|
* close() functions -- what would we do if there was an
|
|
|
|
* error?
|
|
|
|
*/
|
|
|
|
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) ompi_crcp_base_close();
|
|
|
|
#endif
|
|
|
|
(void) ompi_op_base_close();
|
|
|
|
(void) ompi_dpm_base_close();
|
|
|
|
(void) ompi_pubsub_base_close();
|
|
|
|
(void) mca_topo_base_close();
|
|
|
|
(void) mca_btl_base_close();
|
|
|
|
(void) ompi_mtl_base_close();
|
|
|
|
(void) mca_pml_base_close();
|
|
|
|
(void) mca_mpool_base_close();
|
|
|
|
(void) mca_rcache_base_close();
|
|
|
|
(void) mca_io_base_close();
|
2011-09-07 14:23:28 +00:00
|
|
|
(void) mca_fbtl_base_close();
|
|
|
|
(void) mca_fcoll_base_close();
|
|
|
|
(void) mca_fcache_base_close();
|
|
|
|
(void) mca_fs_base_close();
|
|
|
|
(void) mca_sharedfp_base_close();
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) mca_coll_base_close();
|
|
|
|
(void) mca_allocator_base_close();
|
|
|
|
(void) ompi_osc_base_close();
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) orte_grpcomm_base_close();
|
2010-05-03 04:11:03 +00:00
|
|
|
(void) orte_db_base_close();
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) orte_notifier_base_close();
|
|
|
|
(void) orte_ess_base_close();
|
|
|
|
(void) orte_show_help_finalize();
|
|
|
|
#if !ORTE_DISABLE_FULL_SUPPORT
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) orte_snapc_base_close();
|
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
|
|
|
(void) orte_sstore_base_close();
|
2009-06-30 21:41:28 +00:00
|
|
|
#endif
|
|
|
|
(void) orte_filem_base_close();
|
|
|
|
(void) orte_iof_base_close();
|
|
|
|
(void) orte_plm_base_close();
|
|
|
|
(void) orte_odls_base_close();
|
|
|
|
(void) orte_rmaps_base_close();
|
|
|
|
(void) orte_ras_base_close();
|
|
|
|
(void) orte_rml_base_close();
|
|
|
|
(void) orte_routed_base_close();
|
|
|
|
(void) mca_oob_base_close();
|
2009-09-29 13:14:39 +00:00
|
|
|
|
|
|
|
(void) orte_rmcast_base_close();
|
2009-06-30 21:41:28 +00:00
|
|
|
#endif
|
|
|
|
(void) orte_errmgr_base_close();
|
|
|
|
|
|
|
|
(void) opal_backtrace_base_close();
|
|
|
|
(void) opal_memory_base_close();
|
|
|
|
(void) opal_memchecker_base_close();
|
|
|
|
(void) opal_paffinity_base_close();
|
|
|
|
(void) opal_carto_base_close();
|
|
|
|
(void) opal_maffinity_base_close();
|
|
|
|
(void) opal_timer_base_close();
|
2011-09-11 19:02:24 +00:00
|
|
|
#if OPAL_HAVE_HWLOC
|
|
|
|
(void) opal_hwloc_base_close();
|
|
|
|
#endif
|
2010-03-12 23:57:50 +00:00
|
|
|
#if OPAL_ENABLE_FT_CR == 1
|
2009-06-30 21:41:28 +00:00
|
|
|
(void) opal_crs_base_close();
|
|
|
|
#endif
|
Update libevent to the 2.0 series, currently at 2.0.7rc. We will update to their final release when it becomes available. Currently known errors exist in unused portions of the libevent code. This revision passes the IBM test suite on a Linux machine and on a standalone Mac.
This is a fairly intrusive change, but outside of the moving of opal/event to opal/mca/event, the only changes involved (a) changing all calls to opal_event functions to reflect the new framework instead, and (b) ensuring that all opal_event_t objects are properly constructed since they are now true opal_objects.
Note: Shiqing has just returned from vacation and has not yet had a chance to complete the Windows integration. Thus, this commit almost certainly breaks Windows support on the trunk. However, I want this to have a chance to soak for as long as possible before I become less available a week from today (going to be at a class for 5 days, and thus will only be sparingly available) so we can find and fix any problems.
Biggest change is moving the libevent code from opal/event to a new opal/mca/event framework. This was done to make it much easier to update libevent in the future. New versions can be inserted as a new component and tested in parallel with the current version until validated, then we can remove the earlier version if we so choose. This is a statically built framework ala installdirs, so only one component will build at a time. There is no selection logic - the sole compiled component simply loads its function pointers into the opal_event struct.
I have gone thru the code base and converted all the libevent calls I could find. However, I cannot compile nor test every environment. It is therefore quite likely that errors remain in the system. Please keep an eye open for two things:
1. compile-time errors: these will be obvious as calls to the old functions (e.g., opal_evtimer_new) must be replaced by the new framework APIs (e.g., opal_event.evtimer_new)
2. run-time errors: these will likely show up as segfaults due to missing constructors on opal_event_t objects. It appears that it became a typical practice for people to "init" an opal_event_t by simply using memset to zero it out. This will no longer work - you must either OBJ_NEW or OBJ_CONSTRUCT an opal_event_t. I tried to catch these cases, but may have missed some. Believe me, you'll know when you hit it.
There is also the issue of the new libevent "no recursion" behavior. As I described on a recent email, we will have to discuss this and figure out what, if anything, we need to do.
This commit was SVN r23925.
2010-10-24 18:35:54 +00:00
|
|
|
(void) opal_event_base_close();
|
2009-06-30 21:41:28 +00:00
|
|
|
|
|
|
|
/* Do not call OPAL's installdirs close; it will be handled in
|
|
|
|
* opal_finalize_util().
|
|
|
|
*/
|
|
|
|
for (i=0; i < component_map.size; i++) {
|
|
|
|
if (NULL != (map = (ompi_info_component_map_t*)opal_pointer_array_get_item(&component_map, i))) {
|
|
|
|
OBJ_RELEASE(map);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
OBJ_DESTRUCT(&component_map);
|
|
|
|
}
|
|
|
|
|
|
|
|
opened_components = false;
|
|
|
|
}
|