1
1
Граф коммитов

161 Коммитов

Автор SHA1 Сообщение Дата
Ralph Castain
4355134991 Let the vm launcher specify the mapping policy
This commit was SVN r22797.
2010-03-08 19:13:21 +00:00
Ralph Castain
69fe5ca69b Correctly compute bynode mapping, even in the presence of a $#$%#@^$ rankfile
This commit was SVN r22748.
2010-03-02 05:21:42 +00:00
Ralph Castain
5514d9c673 Fix the stupid rankfile mapper again, hopefully not breaking everything else to accommodate it. Looks like the round-robin mappers still work, at least...
This commit was SVN r22746.
2010-03-01 20:40:47 +00:00
Josh Hursey
a3583b8f57 Fix --bynode option to remember for subsequent jobs where it left off last time.
Add a ''map_bynode'' info key to determine if the job to be started by comm_spawn* should be mapped by node or by slot. Default is to map according to the default policy set when the parent job was started.

cmr:v1.5.1

This commit was SVN r22564.
2010-02-05 15:37:49 +00:00
Shiqing Fan
bdc13dacb1 A type cast.
This commit was SVN r22520.
2010-01-31 20:22:22 +00:00
Ralph Castain
7badff9d2d Okay to return no available nodes for mapping when launching daemons - just means there is nothing to do
This commit was SVN r22509.
2010-01-28 22:58:28 +00:00
Ralph Castain
f66b6cae23 Enable the boot of an orted "virtual machine". Modify the mapper framework to allow mapping of only daemons. Remove the cm ras module as no longer required. Modify the orted code to always send back node arch info. Remove the "--enable-bootstrap" configure option as this feature will now always be available.
This commit was SVN r22480.
2010-01-25 22:25:13 +00:00
Ralph Castain
f1f156d57b Make rmaps base open function play nicely with ompi_info
This commit was SVN r22111.
2009-10-20 07:28:23 +00:00
Ralph Castain
d8d80d6f1a Closes trac:2054. Check if a user specifies more cpus-per-rank than there are cpus in a socket - if so, politely tell them "you are stupid" and abort.
This commit was SVN r22091.

The following Trac tickets were found above:
  Ticket 2054 --> https://svn.open-mpi.org/trac/ompi/ticket/2054
2009-10-13 04:19:07 +00:00
Ralph Castain
1475d34c13 Ensure we default to byslot mapping
This commit was SVN r22090.
2009-10-11 23:50:42 +00:00
Ralph Castain
dcab61ad83 Restore the prior default rank assignment scheme for round-robin mappers. Ensure that each app_context has sequential vpids.
This commit was SVN r22048.
2009-10-02 03:16:18 +00:00
Ralph Castain
a15c58c583 Fix the proc assignment into the job data object during assignment of vpids as comm_spawned procs were being overwritten by their parents with the same vpid.
Add a little debug output when updating proc state

This commit was SVN r22042.
2009-10-01 13:44:34 +00:00
Ralph Castain
51f64aaf96 Add a new ras module to support bootstrap operations. Additional functionality may eventually be required in the component, but for now all it does is provide a mechanism for ensuring that other allocations don't confuse the system.
Only active if specifically directed to use it

This commit was SVN r22040.
2009-09-30 23:30:24 +00:00
Ralph Castain
dff0d01673 Yet another paffinity cleanup...sigh.
1. ensure that orte_rmaps_base_schedule_policy does not override cmd line settings

2. when you try to bind to more cores than we have, generate a not-enough-processors error message

3. allow npersocket -bind-to-core combination - because, yes, somebody actually wants to do it.

This commit was SVN r21996.
2009-09-22 18:44:53 +00:00
Ralph Castain
8da3aa8d5c Some (hopefully final!) adjustments and corrections to the paffinity support:
1. default -npersocket to force -bind-to-socket

2. if we cannot get a value for cores/socket, try using #logical cpus. otherwise, default to 1 core

3. add missing error message for not-enough-processors

4. since we no longer loop through orte_register_params twice, put the auto-detect of
   topology info in the rte_init for hnp and std_orted

5. fix bind-to-core, bysocket combination

This commit was SVN r21992.
2009-09-22 15:41:03 +00:00
Ralph Castain
142036f2c0 Issue an error message and abort if the user requests a number of processes that conflicts with nperxxx directives when evaluated against available resources
This commit was SVN r21949.
2009-09-07 03:36:10 +00:00
Ralph Castain
0421a49844 Update the xml support to allow -xml-file foo whereby we redirect all xml formatted output (and ONLY xml formatted output) to a specified file
This commit was SVN r21930.
2009-09-02 18:03:10 +00:00
Ralph Castain
0394a4884d Setup cpus-per-proc and cpus-per-rank as synonyms, both in mca params and on mpirun cmd line
This commit was SVN r21914.
2009-08-30 14:30:36 +00:00
Ralph Castain
2d27bc9824 Default npersocket to bind-to-socket unless otherwise directed
This commit was SVN r21904.
2009-08-27 13:21:14 +00:00
Ralph Castain
5e710928a5 Revise the new binding system slightly:
1. finalize the logic for properly respecting externally assigned bindings. Thanks to Chris Samuel for his help with this. Still needs some acid testing, but appears to now work.

2. remove the double-logic of requiring opal_paffinity_alone AND bind-to-foo. If the user specifies bind-to-foo, trust her and just do it.

This commit was SVN r21885.
2009-08-26 02:01:49 +00:00
Ralph Castain
0005e6e834 Correct a couple of bugs in the rank_file mapper that were incorrectly assigning vpids.
Add a capability to parse the rankfile to extract node information in place of requiring both hostfile and rankfile for non-RM managed environments. The rankfile is -only- parsed for this IF the hostfile and -host options are not given. Otherwise, those are used to establish allocation info as we did before this commit.

This commit was SVN r21815.
2009-08-13 16:08:43 +00:00
Ralph Castain
1dc12046f1 Modify the OMPI paffinity and mapping system to support socket-level mapping and binding. Mostly refactors existing code, with modifications to the odls_default module to support the new capabilities.
Adds several new mpirun options:

* -bysocket - assign ranks on a node by socket. Effectively load balances the procs assigned to a node across the available sockets. Note that ranks can still be bound to a specific core within the socket, or to the entire socket - the mapping is independent of the binding.

* -bind-to-socket - bind each rank to all the cores on the socket to which they are assigned.

* -bind-to-core - currently the default behavior (maintained from prior default)

* -npersocket N - launch N procs for every socket on a node. Note that this implies we know how many sockets are on a node. Mpirun will determine its local values. These can be overridden by provided values, either via MCA param or in a hostfile

Similar features/options are provided at the board level for multi-board nodes.

Documentation to follow...

This commit was SVN r21791.
2009-08-11 02:51:27 +00:00
Ralph Castain
dbac602be5 Add support for the add-host and add-hostfile MPI Info keys to allow Comm_spawn users to add new hosts to those already known by mpirun.
Requires full testing once comm_spawn is fixed (Edgar is working that now).

This commit was SVN r21664.
2009-07-14 14:34:11 +00:00
Ralph Castain
b97f885c00 Restore the original API to terminate individual processes instead of the entire job. This was originally removed as we didn't at that time know how to take advantage of it. Some of us are now working on proactive resilience methods that move procs prior to node failure, so this is now a required API. Modify the odls, plm, and orted functions to support this new functionality.
Continue work on the resilient mapper, completing support for fault groups.

This commit was SVN r21639.
2009-07-13 02:29:17 +00:00
Ralph Castain
b96a71b62e Enable restart of individual processes upon command via the errmgr callback function. It needs an external application to drive this capability, so normal operations shouldn't be affected.
Does not support MPI applications. More work coming to update daemon accounting on movement of procs across nodes.

This commit was SVN r21545.
2009-06-26 20:54:58 +00:00
Ralph Castain
e9fc0a74fb Silence compiler warnings
This commit was SVN r21445.
2009-06-16 13:34:31 +00:00
Ralph Castain
d1dd8c2653 Ensure we accurately count the number of new daemons to be launched, especially if we are restarting processes.
Have the resilient mapper also setup for new daemons in case the PLM needs them.

This commit was SVN r21437.
2009-06-15 13:55:01 +00:00
Ralph Castain
170327e575 Reorg the rmaps components to collect shared code for byslot and bynode mapping in the base so we quit duplicating it in every mapper
This commit was SVN r21424.
2009-06-12 17:52:17 +00:00
Shiqing Fan
5a90b3068e Two type casts.
This commit was SVN r21388.
2009-06-07 12:51:46 +00:00
Ralph Castain
0336460b0a Continue implementation of resilient operations by supporting reuse of jobids for restarted procs. Ensure that restarted processes have valid node and local ranks, and that node rank values are passed to direct-launched processes.
This commit was SVN r21385.
2009-06-06 01:08:47 +00:00
Ralph Castain
5fa3b38d3c Revert r21097 as this results in multiple instantiations of global variables. Instead, fix the problem by including orte_globals.h in the orte_init.c.
Since I already had some changes in there, add in the rmaps rank_file changes - should work okay, but not fully tested.

This commit was SVN r21099.

The following SVN revision numbers were found above:
  r21097 --> open-mpi/ompi@88ae934c26
2009-04-29 02:13:14 +00:00
Rainer Keller
221fb9dbca ... Delayed due to notifier commits earlier this day ...
- Delete unnecessary header files using
   contrib/check_unnecessary_headers.sh after applying
   patches, that include headers, being "lost" due to
   inclusion in one of the now deleted headers...

   In total 817 files are touched.
   In ompi/mpi/c/ header files are moved up into the actual c-file,
   where necessary (these are the only additional #include),
   otherwise it is only deletions of #include (apart from the above
   additions required due to notifier...)

 - To get different MCAs (OpenIB, TM, ALPS), an earlier version was
   successfully compiled (yesterday) on:
   Linux locally using intel-11, gcc-4.3.2 and gcc-SVN + warnings enabled
   Smoky cluster (x86-64 running Linux) using PGI-8.0.2 + warnings enabled
   Lens cluster (x86-64 running Linux) using Pathscale-3.2 + warnings enabled

This commit was SVN r21096.
2009-04-29 01:32:14 +00:00
Rainer Keller
6c1cce8761 - For the upcoming header cleanup commit,
several header files (previously included by header-files)
   now have to be moved "upward".
   This is mainly system headers such as string.h, stdio.h and for
   networking, but also some orte headers.

This commit was SVN r21095.
2009-04-29 00:49:23 +00:00
Rainer Keller
bff1b2a22b - Finally add the missing opal/util/output.h
for the OPAL_OUTPUT_VERBOSE macro.
 - ompi/errhandler/errhandler_predefined.h:
   Well, just the missing fwd declarations...

This commit was SVN r20820.
2009-03-17 22:37:15 +00:00
Rainer Keller
a94438343b - Revert r20740
This commit was SVN r20741.

The following SVN revision numbers were found above:
  r20740 --> open-mpi/ompi@2a70618a77
2009-03-05 21:50:47 +00:00
Rainer Keller
2a70618a77 - Second patch, as discussed in Louisville.
Replace short macros in orte/util/name_fns.h
   to the actual fct. call.

 - Compiles on linux/x86-64

This commit was SVN r20740.
2009-03-05 21:14:18 +00:00
Ralph Castain
f11931306a Modify the accounting system to recycle jobids. Properly recover resources from nodes and jobs upon completion. Adjustments in several places were required to deal with sparsely populated job, node, and proc arrays as a result of this change.
Correct an error wrt how jobids were being computed. Needed to ensure that the job family field was not overrun as we increment jobids for comm_spawn.

Update the slurm plm module so it uses the new slurm termination procedure (brings trunk back into alignment with 1.3 branch).

Update the slurmd ess component so it doesn't get selected if we are running a singleton inside of a slurm allocation.

Cleanup HNP init by moving some code that had been in orte_globals.c for historical reasons into the ess hnp module, and removing the call to that code from the ess_base_std_prolog


NOTE: this change allows orte to support an infinite aggregate number of comm_spawn's, with up to 64k being alive at any one instant. HOWEVER, the MPI layer currently does -not- support re-use of jobids. I did some prototype coding to revise the ompi_proc_t structures, but the BTLs are caching their own data, and there was no readily apparent way to update it. Thus, attempts to spawn more than the 64k limit will abort to avoid causing the MPI layer to hang.

This commit was SVN r20700.
2009-03-03 16:39:13 +00:00
Rainer Keller
04567d3af0 - Header orte/mca/errmgr/errmgr.h is not needed.
Once again compiles fine with -Wimplicit-function-declaration   

This commit was SVN r20640.
2009-02-26 04:05:30 +00:00
Rainer Keller
b356e90fa1 - Get rid of include orte/util/proc_info.h, if not needed
Only proc_info.h-internal include file is opal/dss/dss_types.h
 - In one case (orte/util/hnp_contact.c) had to add proc_info.h again.
 - Local compilation (Linux/x86_64) w/ -Wimplicit-function-declaration
   works fine, no errors.

   Again, let's have MTT the last word.

This commit was SVN r20631.
2009-02-25 03:38:00 +00:00
Rainer Keller
d81443cc5a - On the way to get the BTLs split out and lessen dependency on orte:
Often, orte/util/show_help.h is included, although no functionality
   is required -- instead, most often opal_output.h, or               
   orte/mca/rml/rml_types.h                                           
   Please see orte_show_help_replacement.sh commited next.            

 - Local compilation (Linux/x86_64) w/ -Wimplicit-function-declaration
   actually showed two *missing* #include "orte/util/show_help.h"     
   in orte/mca/odls/base/odls_base_default_fns.c and                  
   in orte/tools/orte-top/orte-top.c                                  
   Manually added these.                                              

   Let's have MTT the last word.

This commit was SVN r20557.
2009-02-14 02:26:12 +00:00
Ralph Castain
89792bbc72 May as well have the other "clean" outputs use the same channel
This commit was SVN r20082.
2008-12-08 19:37:22 +00:00
Ralph Castain
037231fbcb MOdify the node_rank and local_rank fields to be uint16_t so we can handle more than 256 procs/node. Change the type to a defined one so that any future change can be easily done, if required.
This commit was SVN r19637.
2008-09-25 13:39:08 +00:00
Ralph Castain
e64b79f30f Modify the --display-map and --display-alloc per note on devel list to reduce info for user understanding.
Add --display-devel-map and --display-devel-alloc to display all the detailed info we used to provide - it is only of use/interest to developers anyway and confuses users.

This commit was SVN r19608.
2008-09-23 15:46:34 +00:00
Ralph Castain
c0d7fbaf88 A few mapping cleanups - mostly aimed to properly balancing loads so multi app-context comm_spawns don't dump everything on one node.
This commit was SVN r19519.
2008-09-08 15:45:55 +00:00
Ralph Castain
1210a96d82 Ensure a value gets defined before used...thanks Jeff
This commit was SVN r19075.
2008-07-29 13:08:45 +00:00
Ralph Castain
1a77b15523 Modify the handling of hostfiles to allow them to subdivide allocations. Utilize the "slots_alloc" field of the orte_node_t object - which had previously been unused - to track the #slots allocated to a given app_context. Let the hostfile filtering action utilize the #slots field to modify the allocated slots for each app_context.
This commit was SVN r19066.
2008-07-28 15:10:40 +00:00
Ralph Castain
3107545709 Ensure that ORTE processes such as mpirun and orted never inadvertently bind themselves to cores. Change the mca param name used by the rank_file mapper to get user directives on slot lists to be different from that used by MPI procs to discover their binding. Add a cmd line option to orterun to make it easier for a user to specify the slot list (basically, hide the mca param name).
Discussed and reviewed with Lenny and Jeff.

This commit was SVN r19062.
2008-07-28 14:18:36 +00:00
Ralph Castain
718cceddaa Ensure that we only launch procs on the HNP if that node is actually included in the allocation.
This commit was SVN r19038.
2008-07-25 17:13:22 +00:00
Ralph Castain
a1d296ae03 This commit fixes ticket #1410
Fix a few bugs in the mappers:

1. Ensure that bynode with no -np fills all available slots - it just does so with the ranks set bynode instead of byslot

2. fix --nolocal behavior so it works correctly in all cases. We still have to test the host's name using opal_ifislocal in the mapper because the name returned by gethostname to orte_process_info.hostname can be an FQDN, but a hostfile may contain a non-FQDN version.

3. Add missing --nolocal logic to the seq mapper

Oversubscribed mapping seemed to be working okay without repair, so I couldn't verify my own bug report in that regard.

Also included are some preliminary changes to support the modified hostfile behavior, which will be committed shortly:

1. removed the totally useless "allocate" field in the orte_node_t object since every node is automatically allocated for use - and everything ignored the field anyway

2. correctly initialize the slots_alloc field when the allocation is read

This commit was SVN r19030.
2008-07-25 13:35:12 +00:00
Ralph Castain
0532d799d6 Complete implementation of the --without-rte-support configure option. Working with Brian, this has been tested on RedStorm.
Some minor changes to help facilitate debugger support so that both mpirun and yod can operate with it. Still to be completed.

This commit was SVN r18664.
2008-06-18 03:15:56 +00:00