ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
16 089 коммитов 3 Ветки 0 Теги 130 MiB
5711414eb7
Граф коммитов

2494 Коммитов

Автор SHA1 Сообщение Дата
Shiqing Fan
5711414eb7 Fix Windows build 
This commit was SVN r25351.
 2011-10-21 14:46:58 +00:00
Ralph Castain
53ef085567 Fix a minor issue seen by Jeff in specific failure pathway 
This commit was SVN r25350.
 2011-10-21 14:44:48 +00:00
Ralph Castain
3e72fccacf Cray's PMI implementation is quite different from slurm's - they extended PMI-1 by adding some, but not all, of the PMI-2 APIs. So you can't just switch to using PMI-2 functions as it isn't a complete implementation. Instead, you have to selectively figure out which ones they have in PMI-2, and use any missing ones from PMI-1. What fun. 
Modify the configure logic and the PMI components to accommodate Cray's approach. Refactor the PMI error reporting code so it resides in only one place. Cray actually decided -not- to define the PMI-2 error codes, so we have to use the PMI-1 codes instead. More fun.

This commit was SVN r25348.
 2011-10-21 04:54:38 +00:00
Nathan Hjelm
beb8d8ce32 pmi return code wtf 
This commit was SVN r25336.
 2011-10-20 17:51:24 +00:00
Ralph Castain
84713d5a84 Fix singletons again - must have been broken for a very long time, which only shows how little anyone cares about this capability. 
This commit was SVN r25332.
 2011-10-19 20:19:08 +00:00
Ralph Castain
b44f8d4b28 Complete implementation of the ess.proc_get_locality API. Up to this point, the API was only capable of telling if the specified proc was sharing a node with you. However, the returned value was capable of telling you much more detailed info - e.g., if the proc shares a socket, a cache, or numa node. We just didn't have the data to provide that detail. 
Use hwloc to obtain the cpuset for each process during mpi_init, and share that info in the modex. As it arrives, use a new opal_hwloc_base utility function to parse the value against the local proc's cpuset and determine where they overlap. Cache the value in the pmap object as it may be referenced multiple times.

Thus, the return value from orte_ess.proc_get_locality is a 16-bit bitmask that describes the resources being shared with you. This bitmask can be tested using the macros in opal/mca/paffinity/paffinity.h

Locality is available for all procs, whether launched via mpirun or directly with an external launcher such as slurm or aprun.

This commit was SVN r25331.
 2011-10-19 20:18:14 +00:00
Ralph Castain
2958f3de34 Add some clarifying comments and a small efficiency improvement 
This commit was SVN r25322.
 2011-10-18 18:30:43 +00:00
Ralph Castain
b771114086 Fix the fix :-) 
If the errmgr is going to try and hold the orted until all routes and children are gone, then the exit cmd must do the same. Otherwise, the orted exits immediately without waiting for routes to be dismantled, which is why we don't see the connections close.

Also cleanup some diagnostics and add some debug to more clearly see what's going on.

This commit was SVN r25321.
 2011-10-18 17:56:37 +00:00
Ralph Castain
ae8e556d14 Okay, once again let's fix the vpid calculator. Identified problem with prior commit (some rmaps components already place their procs in the jdata->procs array, and others don't), so account for those variations. 
This commit was SVN r25315.
 2011-10-18 15:50:11 +00:00
George Bosilca
749b63c09d Provide a generic fix for the termination issue instead of r25248. The 
termination condition is to be checked at the daemon/HNP level not down
in the routing.

This commit was SVN r25313.

The following SVN revision numbers were found above:
  r25248 --> open-mpi/ompi@b42ccc89b8
 2011-10-18 03:07:37 +00:00
George Bosilca
f28890fbb7 Revert r25302 as it break the --bynode option. 
This commit was SVN r25311.

The following SVN revision numbers were found above:
  r25302 --> open-mpi/ompi@d7a8553179
 2011-10-18 02:48:17 +00:00
Ralph Castain
2fdd9c6dea Ensure mpirun doesn't pick this component 
This commit was SVN r25307.
 2011-10-17 22:28:28 +00:00
Ralph Castain
8f0ef54130 Complete implementation of pmi support. Ensure we support both mpirun and direct launch within same configuration to avoid requiring separate builds. Add support for generic pmi, not just under slurm. Add publish/subscribe support, although slurm's pmi implementation will just return an error as it hasn't been done yet. 
This commit was SVN r25303.
 2011-10-17 20:51:22 +00:00
Ralph Castain
d7a8553179 Fix the mapping algo for computing vpids - it was borked for bynode operations when using nperxxx directives 
This commit was SVN r25302.
 2011-10-17 19:49:04 +00:00
Ralph Castain
89a20de474 Remove unused includes. Ensure that the error log is at least always available as we otherwise segfault when reporting errors that occur prior to opening the errmgr framework 
This commit was SVN r25288.
 2011-10-14 18:45:11 +00:00
Ralph Castain
07dbbc6513 Sorry for mid-day correction - but folks are trying to test this, and we didn't realize it was still ignored :-( 
This commit was SVN r25287.
 2011-10-14 16:19:20 +00:00
Ralph Castain
054c485dcf Cleanup a race condition and an unreliable method that caused us to not properly handle procs that trapped sigterm for cleanup purposes while ORTE was trying to kill them. Thanks to Rick Payne and Ian Wells of Cisco for spending weeks chasing this down. 
Fix a termination issue that caused procs local to mpirun to not be killed if they weren't calling into the library. Thanks to Terry Dontje for spending countless hours chasing his tail on this one! :-(

This commit was SVN r25285.
 2011-10-14 15:39:54 +00:00
Ralph Castain
08fa9e1c6a Correct include path 
This commit was SVN r25282.
 2011-10-13 23:46:52 +00:00
Ralph Castain
b96ef2161d Complete the PMI support. Generalize PMI operations to support both slurm and non-slurm environments. Correct some configuration issues - we really only want the PMI integration at the individual component level. Ensure that the pmi grpcomm component doesn't get selected when launching via mpirun by setting its priority below the bad component. 
Only verified in a slurm environment as that's all I have access to...

This commit was SVN r25275.
 2011-10-12 20:59:25 +00:00
Ralph Castain
634f83fc52 Fix the routed components. All had errors, some completely broken. You cannot test 
0 == ORTE_EPOCH_CMP(target->epoch,ORTE_EPOCH_INVALID)

when epoch is not configured as this will always return true. This caused get_route to return an error in all non-binomial routed modules, and caused all components to return an error when delete_route was called.

So protect the checks with ORTE_ENABLE_EPOCH so we get the correct behavior.

This commit was SVN r25274.
 2011-10-12 20:18:57 +00:00
Ralph Castain
24a46f2acb These were missed by prior commit - need to remove lingering references to OPAL_HWLOC_HAVE_XML 
This commit was SVN r25272.
 2011-10-12 16:54:03 +00:00
George Bosilca
872d377021 Tell what the update status is. 
This commit was SVN r25259.
 2011-10-11 19:49:12 +00:00
Brian Barrett
98e98ce2c5 * opal_atomic_trylock is documented to return 0 if the lock was acquired, 
1 otherwise.  It was doing the opposite, so this patch fixes the
  return values.  All uses (all in ORTE) used the actual return values,
  not the documented values, so fix them as well.

This commit was SVN r25257.
 2011-10-11 18:43:45 +00:00
Ralph Castain
2f38ff5e54 Ensure we don't try to build this module unless pmi is specifically requested 
This commit was SVN r25252.
 2011-10-11 06:12:04 +00:00
Ralph Castain
baefdabd98 Add some debug. Now confirmed to work correctly (prior problem was with odin tcp connection, not code). 
This commit was SVN r25249.
 2011-10-11 02:15:17 +00:00
Ralph Castain
b42ccc89b8 Although this didn't solve the earlier termination problem, the code will be required once we get connection terminations properly detected. If a daemon (or HNP) is trying to terminate, then we need to check for termination conditions whenever a route is lost - when all child connections are gone, then we are free to finalize. 
This commit was SVN r25248.
 2011-10-10 21:41:49 +00:00
Ralph Castain
1aa1c2e9b4 Get the slurm pmi support working. Cannot use infiniband, of course, as the oob can't make the connection - may try other existing methods. Modex may not quite be working right yet 
as odin was having trouble making TCP connections, but at least the configure now works so things build, so save that for now

This commit was SVN r25247.
 2011-10-10 21:39:10 +00:00
Swen Boehm
08b4322a1a patched the lex files to not issue the following compiler warning: 
'yyunput' defined but not used

This commit was SVN r25246.
 2011-10-10 18:13:04 +00:00
George Bosilca
649af6c925 Enumerated mixed with another type (int) is tolerated but 
easily fixable.

This commit was SVN r25241.
 2011-10-09 03:54:52 +00:00
Terry Dontje
c6691b4122 clean up local procs when abort or abort signal happens 
This commit was SVN r25237.
 2011-10-06 19:19:55 +00:00
George Bosilca
c6d6c9aece Remove some #if by using the correct macro (aka. ORTE_EPOCH_CMP). 
This commit was SVN r25229.
 2011-10-04 14:42:40 +00:00
Samuel Gutierrez
25cbf79592 modifications to ras alps. this commit allows users to mpirun without having to set id environment variables (BASIL_RESERVATION_ID, OMPI_ALPS_RESID). note, however, that we preserved the old behavior. if an id environment variable is set, it will be obeyed and our new code path is essentially bypassed. if we missed something, please yell at us. with this commit, the use of ras-alps-command.sh is no longer needed... at least that is our hope. 
This commit was SVN r25181.
 2011-09-26 21:31:08 +00:00
Ralph Castain
8347385630 Fix the radix routed component. 
This commit was SVN r25175.
 2011-09-22 09:32:53 +00:00
Ralph Castain
1cd7b02df3 Add a set of default errmgr components that support solely the default "everything dies on error" behavior. Set their priority to be selected by default, but provide params to adjust those priorities to allow other component selection. 
This commit was SVN r25139.
 2011-09-13 22:03:45 +00:00
Ralph Castain
3c4f04f4d9 Ensure opal_hwloc_topology is NULL after being destroyed 
This commit was SVN r25138.
 2011-09-13 19:21:10 +00:00
Nathan Hjelm
079ccdf8b1 fix debugger co-location launching 
This commit was SVN r25136.
 2011-09-13 15:08:03 +00:00
Ralph Castain
ca7638553f Remove stale code 
This commit was SVN r25133.
 2011-09-12 23:00:41 +00:00
Ralph Castain
556a05566e Silence warning 
This commit was SVN r25130.
 2011-09-12 16:21:51 +00:00
Ralph Castain
92c7372e20 Per the RFC from Jeff, move hwloc from opal/mca/common to its own static framework ala libevent. Have ORTE daemons collect the topology info at startup and, if --enable-hwloc-xml is set, send that info back to the HNP for later use. The HNP only retains unique topology "templates" to reduce memory footprint. Have the daemon include the local topology info in the nidmap buffer sent to each app so the apps don't all hammer the local system to discover it for themselves. 
Remove the sysinfo framework as hwloc replaces that functionality.

This commit was SVN r25124.
 2011-09-11 19:02:24 +00:00
Ralph Castain
2091e39bee Record the file descriptor on the read event when building optimized 
This commit was SVN r25123.
 2011-09-11 18:57:14 +00:00
Wesley Bland
f8740e5478 Correct a typo reported by Pasha. 
This commit was SVN r25109.
 2011-08-30 18:44:52 +00:00
Ralph Castain
03ddf8520b Resolve not-used warnings 
This commit was SVN r25101.
 2011-08-27 14:27:15 +00:00
Wesley Bland
f542ecd578 Fix a couple of problems with the resil code not compiling. 
This commit was SVN r25099.
 2011-08-27 03:21:00 +00:00
George Bosilca
a4245b8d63 Remove some warnings related to the resilience patch. 
This commit was SVN r25097.
 2011-08-27 00:15:34 +00:00
George Bosilca
67ec5a0556 While doing cleanup remove pending warnings. 
This commit was SVN r25095.
 2011-08-26 23:38:06 +00:00
Wesley Bland
4e7ff0bd5e By popular demand the epoch code is now disabled by default. 
To enable the epochs and the resilient orte code, use the configure flag:

--enable-resilient-orte

This will define both:

ORTE_ENABLE_EPOCH
ORTE_RESIL_ORTE

This commit was SVN r25093.
 2011-08-26 22:16:14 +00:00
Ralph Castain
1c08a4006c Refactor some code to remove a few API handles from errmgr. Reviewed/tested by Wes. 
This commit was SVN r25064.
 2011-08-18 16:24:45 +00:00
Ralph Castain
e58623cd5b Bring alps back to full operations by correctly computing daemon names. Unfortunately, alps doesn't assign cnos rank in node-based order - i.e., cnos rank=0 isn't necessarily on the first node of the execution. So adjust when using static ports. 
Add some debug to nidmap

Ensure that the HNP's node name is not included in the regex when launching via rshbase as that node is automatically included in the daemon map.

This commit was SVN r25063.
 2011-08-18 14:59:18 +00:00
Wesley Bland
a2a20c3766 I believe this should fix the race condition that Terry is seeing in the MTT 
tests. It appears that nothing in the errmgr was using the mutexes to protect
the odls child list.

This commit was SVN r25062.
 2011-08-18 14:52:30 +00:00
Ralph Castain
23f47295a8 Add even more debug 
This commit was SVN r25053.
 2011-08-16 16:41:33 +00:00