1
1
Граф коммитов

58 Коммитов

Автор SHA1 Сообщение Дата
Ralph Castain
6ee32767d4 Restore the cpus-per-proc option for byslot and bynode mapping. Remove the bind_idx (which recorded the index of the hwloc object where the proc was bound) as this would no longer be unique, and just use the bitmap as the standard reference for location. Update the relative locality computation to take bitmaps as its argument.
This commit was SVN r28219.
2013-03-26 18:27:50 +00:00
Ralph Castain
81d0b06842 Strip the domain info from the hostname if that option is specified, protecting IP address-based names
This commit was SVN r27586.
2012-11-10 14:05:27 +00:00
Ralph Castain
3938ec5361 Remove debug
This commit was SVN r27024.
2012-08-13 21:35:21 +00:00
Ralph Castain
b9b41d8662 For cases where the alpha+non-zero prefix must be removed from a node name, be sure to do it everywhere we access node names - otherwise, modex methods such as pmi will fail to correctly identify procs on the same node
This commit was SVN r27022.
2012-08-13 20:44:56 +00:00
Ralph Castain
6ee35e4977 Add num_local_peers to orte_process_info so we don't keep re-computing it, ensure it is available for direct launch via pmi as well
This commit was SVN r26931.
2012-07-31 21:21:50 +00:00
Ralph Castain
0dfe29b1a6 Roll in the rest of the modex change. Eliminate all non-modex API access of RTE info from the MPI layer - in some cases, the info was already present (either in the ompi_proc_t or in the orte_process_info struct) and no call was necessary. This removes all calls to orte_ess from the MPI layer. Calls to orte_grpcomm remain required.
Update all the orte ess components to remove their associated APIs for retrieving proc data. Update the grpcomm API to reflect transfer of set/get modex info to the db framework.

Note that this doesn't recreate the old GPR. This is strictly a local db storage that may (at some point) obtain any missing data from the local daemon as part of an async methodology. The framework allows us to experiment with such methods without perturbing the default one.

This commit was SVN r26678.
2012-06-27 14:53:55 +00:00
Ralph Castain
c69a04e16b Cleanup the pidmap decoding for apps to avoid confusion
This commit was SVN r26498.
2012-05-27 16:21:38 +00:00
Ralph Castain
bd8b4f7f1e Sorry for mid-day commit, but I had promised on the call to do this upon my return.
Roll in the ORTE state machine. Remove last traces of opal_sos. Remove UTK epoch code.

Please see the various emails about the state machine change for details. I'll send something out later with more info on the new arch.

This commit was SVN r26242.
2012-04-06 14:23:13 +00:00
Jeff Squyres
6fbbfd0f7a Gah! r25545 acidentally included ''waaaay'' more stuff than it was
supposed to.  I.e., half-baked/not complete stuff.

This commit backs out all of r25545.  Sorry folks!

This commit was SVN r25546.

The following SVN revision numbers were found above:
  r25545 --> open-mpi/ompi@7f9ae11faf
2011-11-29 23:24:52 +00:00
Jeff Squyres
7f9ae11faf Per http://www.open-mpi.org/community/lists/users/2011/11/17862.php,
to make MPI_IN_PLACE (and other sentinel Fortran constants) work on OS
X, we need to use the following compiler (linker) flag:

    -Wl,-commons,use_dylibs 

So if we're compiling on OS X, test to see if that flag works with the
compiler.  If so, add it to the wrapper FFLAGS and FCFLAGS (note that
per a future update, we'll only have one Fortran compiler anyway).

Fixes trac:1982.  

This commit was SVN r25545.

The following Trac tickets were found above:
  Ticket 1982 --> https://svn.open-mpi.org/trac/ompi/ticket/1982
2011-11-29 23:05:54 +00:00
Ralph Castain
6310361532 At long last, the fabled revision to the affinity system has arrived. A more detailed explanation of how this all works will be presented here:
https://svn.open-mpi.org/trac/ompi/wiki/ProcessPlacement

The wiki page is incomplete at the moment, but I hope to complete it over the next few days. I will provide updates on the devel list. As the wiki page states, the default and most commonly used options remain unchanged (except as noted below). New, esoteric and complex options have been added, but unless you are a true masochist, you are unlikely to use many of them beyond perhaps an initial curiosity-motivated experimentation.

In a nutshell, this commit revamps the map/rank/bind procedure to take into account topology info on the compute nodes. I have, for the most part, preserved the default behaviors, with three notable exceptions:

1. I have at long last bowed my head in submission to the system admin's of managed clusters. For years, they have complained about our default of allowing users to oversubscribe nodes - i.e., to run more processes on a node than allocated slots. Accordingly, I have modified the default behavior: if you are running off of hostfile/dash-host allocated nodes, then the default is to allow oversubscription. If you are running off of RM-allocated nodes, then the default is to NOT allow oversubscription. Flags to override these behaviors are provided, so this only affects the default behavior.

2. both cpus/rank and stride have been removed. The latter was demanded by those who didn't understand the purpose behind it - and I agreed as the users who requested it are no longer using it. The former was removed temporarily pending implementation.

3. vm launch is now the sole method for starting OMPI. It was just too darned hard to maintain multiple launch procedures - maybe someday, provided someone can demonstrate a reason to do so.

As Jeff stated, it is impossible to fully test a change of this size. I have tested it on Linux and Mac, covering all the default and simple options, singletons, and comm_spawn. That said, I'm sure others will find problems, so I'll be watching MTT results until this stabilizes.

This commit was SVN r25476.
2011-11-15 03:40:11 +00:00
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
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
Wesley Bland
84be81df95 Standardize the initialization of the EPOCH's.
Everyone will be starting at MIN anyway (until we implement restart of course)
so there's no reason to set the epoch to INVALID and then immediately reset them
to MIN. This way there's less room to make mistakes later.

This commit was SVN r24829.
2011-06-28 14:20:33 +00:00
Ralph Castain
c203eee223 Since process names now have three fields, be sure to initialize all three of them
This commit was SVN r24828.
2011-06-27 20:50:08 +00:00
Wesley Bland
e1ba09ad51 Add a resilience to ORTE. Allows the runtime to continue after a process (or
ORTED) failure. Note that more work will be necessary to allow the MPI layer to
take advantage of this.

Per RFC:
http://www.open-mpi.org/community/lists/devel/2011/06/9299.php

This commit was SVN r24815.
2011-06-23 20:38:02 +00:00
Ralph Castain
042ee3ec48 Support the option of outputting error_log messages with something other than the process name
This commit was SVN r24784.
2011-06-17 14:50:00 +00:00
Ralph Castain
3ca0b4138b Let the nidmap functions update a new orte_process_info field as to the number of daemons in the system
This commit was SVN r23088.
2010-05-04 02:40:09 +00:00
Ralph Castain
2541aa98ab Change the app_idx type to uint32_t to support users who use large numbers of app_contexts. Set it up as a new typedef so we can change it later without as much effort.
This commit was SVN r22727.
2010-02-27 17:37:34 +00:00
Ralph Castain
9f3ccebeaa We need to barrier for orte apps when the job is initially started, but we must not do the barrier when a proc is restarted as the other procs in the job won't know to participate.
This commit was SVN r22388.
2010-01-10 02:21:30 +00:00
Ralph Castain
60edbc7220 Fix hetero operations and comm_spawn (to a point).
Remove all architecture references from ORTE and put them back in the modex using modex_send/recv calls.

Hetero operations are now fully supported again. Comm_spawn now works up to the point where it segfaults due to an error in the CID code - which now allows Edgar to dig further! :-)

This commit was SVN r21655.
2009-07-13 20:03:41 +00:00
Ralph Castain
4be24521aa Modify the orte_process_info structure to handle a broader range of process types by replacing the individual booleans with a 32-bit bitmap. Use a set of #define's to define the individual bits, and a set of matching macros to test for them. Update the orte code base to use the macros instead of the booleans.
Minor mod to the ompi layer to use the new #define's - just one-line name replacements.

This commit was SVN r21144.
2009-05-04 11:07:40 +00:00
Rainer Keller
ec0ed48718 - Revert r20739
This commit was SVN r20742.

The following SVN revision numbers were found above:
  r20739 --> open-mpi/ompi@781caee0b6
2009-03-05 21:56:03 +00:00
Rainer Keller
781caee0b6 - First of two or three patches, in orte/util/proc_info.h:
Adapt orte_process_info to orte_proc_info, and
   change orte_proc_info() to orte_proc_info_init().
 - Compiled on linux-x86-64
 - Discussed with Ralph

This commit was SVN r20739.
2009-03-05 20:36:44 +00:00
Ralph Castain
6db641c86d Pass the number of nodes in a job to the process
This commit was SVN r20595.
2009-02-19 20:45:07 +00:00
Jeff Squyres
91d302fd67 A bunch of minor ORTE valgrind-inspired memory leak cleanups (reviewed
by Ralph).

This commit was SVN r20544.
2009-02-13 04:14:10 +00:00
Ralph Castain
631d7d2a85 Handle cases where daemon uri has quote marks around it
This commit was SVN r20491.
2009-02-09 20:40:17 +00:00
Ralph Castain
2966206f58 Fix a race condition in the IOF and add some new user-requested features:
1. fix a race condition whereby a proc's output could trigger an event prior to the other outputs being setup, thus c ausing the IOF to declare the proc "terminated" too early. This was really rare, but could happen.

2. add a new "timestamp-output" option that timestamp's each line of output

3. add a new "output-filename" option that redirects each proc's output to a separate rank-named file.

4. add a new "xterm" option that redirects the output of the specified ranks to a separate xterm window.

This commit was SVN r20392.
2009-01-30 22:47:30 +00:00
Ralph Castain
5e6d3ba289 Initial implementation of static ports. Provide an mca param to specify static port ranges to the OOB - can provide an
y combination of comma-separated values and ranges. Daemons will use the first port in the range, MPI procs will use the other ports in the range assuming that they know their node rank in time and enough ports were specified.

NOTE: this capability only works under specific conditions. I will outline more about this in a note to devel as the remainder of the implementation progresses. For now, the only environment where this works is slurm. The linear routed module has also been adjusted to work with static ports so that all messaging flows strictly through the topology, including the initial daemon callback - thus limiting the number of sockets opened by mpirun.

This commit was SVN r20390.
2009-01-30 18:31:43 +00:00
Ralph Castain
ba5498cdc6 Repair the MPI-2 dynamic operations. This includes:
1. repair of the linear and direct routed modules

2. repair of the ompi/pubsub/orte module to correctly init routes to the ompi-server, and correctly handle failure to correctly parse the provided ompi-server URI

3. modification of orterun to accept both "file" and "FILE" for designating where the ompi-server URI is to be found - purely a convenience feature

4. resolution of a message ordering problem during the connect/accept handshake that allowed the "send-first" proc to attempt to send to the "recv-first" proc before the HNP had actually updated its routes.

Let this be a further reminder to all - message ordering is NOT guaranteed in the OOB

5. Repair the ompi/dpm/orte module to correctly init routes during connect/accept.

Reminder to all: messages sent to procs in another job family (i.e., started by a different mpirun) are ALWAYS routed through the respective HNPs. As per the comments in orte/routed, this is REQUIRED to maintain connect/accept (where only the root proc on each side is capable of init'ing the routes), allow communication between mpirun's using different routing modules, and to minimize connections on tools such as ompi-server. It is all taken care of "under the covers" by the OOB to ensure that a route back to the sender is maintained, even when the different mpirun's are using different routed modules.

6. corrections in the orte/odls to ensure proper identification of daemons participating in a dynamic launch

7. corrections in build/nidmap to support update of an existing nidmap during dynamic launch

8. corrected implementation of the update_arch function in the ESS, along with consolidation of a number of ESS operations into base functions for easier maintenance. The ability to support info from multiple jobs was added, although we don't currently do so - this will come later to support further fault recovery strategies

9. minor updates to several functions to remove unnecessary and/or no longer used variables and envar's, add some debugging output, etc.

10. addition of a new macro ORTE_PROC_IS_DAEMON that resolves to true if the provided proc is a daemon

There is still more cleanup to be done for efficiency, but this at least works.

Tested on single-node Mac, multi-node SLURM via odin. Tests included connect/accept, publish/lookup/unpublish, comm_spawn, comm_spawn_multiple, and singleton comm_spawn.

Fixes ticket #1256

This commit was SVN r18804.
2008-07-03 17:53:37 +00:00
George Bosilca
0f9b9c0aff Remove a warning and add arequired header (otherwise we cannot compile when
--disable-debug is specified).

This commit was SVN r18665.
2008-06-18 08:10:02 +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
Ralph Castain
9613b3176c Effectively revert the orte_output system and return to direct use of opal_output at all levels. Retain the orte_show_help subsystem to allow aggregation of show_help messages at the HNP.
After much work by Jeff and myself, and quite a lot of discussion, it has become clear that we simply cannot resolve the infinite loops caused by RML-involved subsystems calling orte_output. The original rationale for the change to orte_output has also been reduced by shifting the output of XML-formatted vs human readable messages to an alternative approach.

I have globally replaced the orte_output/ORTE_OUTPUT calls in the code base, as well as the corresponding .h file name. I have test compiled and run this on the various environments within my reach, so hopefully this will prove minimally disruptive.

This commit was SVN r18619.
2008-06-09 14:53:58 +00:00
Ralph Castain
828ae26d90 ORTE-level MCA params are defined in several places. Ompi_info cannot call orte_init due to an issue with the memory allocator, thus making it impossible for ompi_info to display all of the ORTE-level MCA params.
By consolidating them all into one function, ompi_info can call that function and register the desired variables. This also requires, however, that ompi_info call orte_output_init to avoid generating tons of error messages, so make that adjustment too. 

Fixes ticket #1314

In addition, orte_output has a race condition issue whereby calls to orte_output/verbose can occur prior to either the RML being defined/setup, or the HNP being defined. This latter occurs during the initialization of the orte_process_info structure. In both cases, there is no way orte_output can send the output to the HNP. Hence, the message must be simply output locally.

Fixes ticket #1315

This commit was SVN r18524.
2008-05-28 13:29:58 +00:00
Jeff Squyres
e7ecd56bd2 This commit represents a bunch of work on a Mercurial side branch. As
such, the commit message back to the master SVN repository is fairly
long.

= ORTE Job-Level Output Messages =

Add two new interfaces that should be used for all new code throughout
the ORTE and OMPI layers (we already make the search-and-replace on
the existing ORTE / OMPI layers):

 * orte_output(): (and corresponding friends ORTE_OUTPUT,
   orte_output_verbose, etc.)  This function sends the output directly
   to the HNP for processing as part of a job-specific output
   channel.  It supports all the same outputs as opal_output()
   (syslog, file, stdout, stderr), but for stdout/stderr, the output
   is sent to the HNP for processing and output.  More on this below.
 * orte_show_help(): This function is a drop-in-replacement for
   opal_show_help(), with two differences in functionality:
   1. the rendered text help message output is sent to the HNP for
      display (rather than outputting directly into the process' stderr
      stream)
   1. the HNP detects duplicate help messages and does not display them
      (so that you don't see the same error message N times, once from
      each of your N MPI processes); instead, it counts "new" instances
      of the help message and displays a message every ~5 seconds when
      there are new ones ("I got X new copies of the help message...")

opal_show_help and opal_output still exist, but they only output in
the current process.  The intent for the new orte_* functions is that
they can apply job-level intelligence to the output.  As such, we
recommend that all new ORTE and OMPI code use the new orte_*
functions, not thei opal_* functions.

=== New code ===

For ORTE and OMPI programmers, here's what you need to do differently
in new code:

 * Do not include opal/util/show_help.h or opal/util/output.h.
   Instead, include orte/util/output.h (this one header file has
   declarations for both the orte_output() series of functions and
   orte_show_help()).
 * Effectively s/opal_output/orte_output/gi throughout your code.
   Note that orte_output_open() takes a slightly different argument
   list (as a way to pass data to the filtering stream -- see below),
   so you if explicitly call opal_output_open(), you'll need to
   slightly adapt to the new signature of orte_output_open().
 * Literally s/opal_show_help/orte_show_help/.  The function signature
   is identical.

=== Notes ===

 * orte_output'ing to stream 0 will do similar to what
   opal_output'ing did, so leaving a hard-coded "0" as the first
   argument is safe.
 * For systems that do not use ORTE's RML or the HNP, the effect of
   orte_output_* and orte_show_help will be identical to their opal
   counterparts (the additional information passed to
   orte_output_open() will be lost!).  Indeed, the orte_* functions
   simply become trivial wrappers to their opal_* counterparts.  Note
   that we have not tested this; the code is simple but it is quite
   possible that we mucked something up.

= Filter Framework =

Messages sent view the new orte_* functions described above and
messages output via the IOF on the HNP will now optionally be passed
through a new "filter" framework before being output to
stdout/stderr.  The "filter" OPAL MCA framework is intended to allow
preprocessing to messages before they are sent to their final
destinations.  The first component that was written in the filter
framework was to create an XML stream, segregating all the messages
into different XML tags, etc.  This will allow 3rd party tools to read
the stdout/stderr from the HNP and be able to know exactly what each
text message is (e.g., a help message, another OMPI infrastructure
message, stdout from the user process, stderr from the user process,
etc.).

Filtering is not active by default.  Filter components must be
specifically requested, such as:

{{{
$ mpirun --mca filter xml ...
}}}

There can only be one filter component active.

= New MCA Parameters =

The new functionality described above introduces two new MCA
parameters:

 * '''orte_base_help_aggregate''': Defaults to 1 (true), meaning that
   help messages will be aggregated, as described above.  If set to 0,
   all help messages will be displayed, even if they are duplicates
   (i.e., the original behavior).
 * '''orte_base_show_output_recursions''': An MCA parameter to help
   debug one of the known issues, described below.  It is likely that
   this MCA parameter will disappear before v1.3 final.

= Known Issues =

 * The XML filter component is not complete.  The current output from
   this component is preliminary and not real XML.  A bit more work
   needs to be done to configure.m4 search for an appropriate XML
   library/link it in/use it at run time.
 * There are possible recursion loops in the orte_output() and
   orte_show_help() functions -- e.g., if RML send calls orte_output()
   or orte_show_help().  We have some ideas how to fix these, but
   figured that it was ok to commit before feature freeze with known
   issues.  The code currently contains sub-optimal workarounds so
   that this will not be a problem, but it would be good to actually
   solve the problem rather than have hackish workarounds before v1.3 final.

This commit was SVN r18434.
2008-05-13 20:00:55 +00:00
Josh Hursey
dcd21d7d07 Some checkpoint/restart fixes in response to r18338 (changes in modex).
Things should be working now.

This commit was SVN r18348.

The following SVN revision numbers were found above:
  r18338 --> open-mpi/ompi@3e55fe6f6d
2008-05-01 17:48:13 +00:00
Ralph Castain
3e55fe6f6d Fold in the revised modex scheme. Move the ompi_proc_t modex portions to the RTE level since the daemons already have that info. Provide each process with the equivalent of a "nidmap" - both a map of what nodes are in the job, and a map of which node each process is on. This enables the use of static ports, though that hasn't been turned "on" in this commit.
Update the rsh tree spawn capability so we spawn the next wave of daemons before launching our own local procs.

Add an ability to encode nodenames for large clusters with contiguous node name numbering schemes - this allows communication of all node names in a few bytes instead of tens-of-bytes/node.

This commit was SVN r18338.
2008-04-30 19:49:53 +00:00
Josh Hursey
cc83d41ad9 Merge in tmp/jjh-scratch
{{{
 svn merge -r 18218:18240 https://svn.open-mpi.org/svn/ompi/tmp/jjh-scratch .
}}}

Contains:
 * Primarily a fix for a user reported problem where a cached file descriptor is causing a SIGPIPE on restart.
 * Cleanup some small memory leaks from using mca_base_param_env_var() - Thanks Jeff
 * Cleanup ORTE FT tool compilation in non-FT builds - Thanks Tim P.
 * Cleanup mpi interface with missplaced {{{OPAL_CR_ENTER_LIBRARY}}} - Thanks Terry
 * Some other sundry cleanup items all dealing with C/R functionality in the trunk.

This commit was SVN r18241.
2008-04-23 00:17:12 +00:00
Ralph Castain
e050f37578 Cleanup a few warnings about initializing variables.
Remove an obsolete data value.

This commit was SVN r18129.
2008-04-10 19:15:16 +00:00
Ralph Castain
537395b924 Make two important MCA params "visible" to ompi_info
This commit was SVN r18074.
2008-04-02 14:54:57 +00:00
Ralph Castain
6166278e18 Improve the scalability of the modex operation and fix a bug reported by Tim P
The bug was a race condition in the barrier operation that caused the barrier in MPI_Finalize to fail on very short programs.

Scalaiblity was improved by using the daemons to aggregate modex and barrier messages before sending them to the rank=0 proc. Improvement is proportional to ppn, of course, but there really wasn't a scaling problem at low ppn anyway. This modification also paves the way for better allgather operations since now all the data for each node is sitting at the daemon level, and the daemons are now aware that a collective operation on the OOB is underway (so they -can- participate in a collective of their own to support it).

Also added better diagnostics to map out the timing associated with MPI_Init - turned on by -mca orte_timing 1.

This commit was SVN r17988.
2008-03-27 15:17:53 +00:00
Ralph Castain
60d931217f Modify the routed framework to allow greater control/flexibility over response to lost routes and initial wireup of jobs as required by several soon-to-come new modules.
Specifically, add two new APIs:

1. lost_route: allows the OOB to report that a connection has failed, thereby giving the routed module an opportunity to respond appropriately to its topology. Creating the API also allows each routed component to hold its own definition of "lifeline" - in some cases, this may be a single connection, but in others it may be multiple connections. Some modules may choose to re-route messaging if the lifeline or any other connection is lost, while others may choose to abort the job.

Both the tree and unity modules retain the current behavior and abort the job if the lifeline connection is lost, while ignoring other lost connections.

2. get_wireup_info: returns (in a provided buffer) info required to wireup connections for the specified job. Some routed modules do not need to return any info as they can wireup via alternative means, while some need to xchg data with their peers. If info is inserted into the buffer, the plm_base_launch_apps function will xcast the contents to the specified job.

The commit also removes the "lifeline" entry from the orte_process_info struct (and the associated ORTE_PROC_MY_LIFELINE definition) as the lifeline info is now contained within the respective routed module.

This commit was SVN r17969.
2008-03-26 01:00:24 +00:00
Ralph Castain
ec76fe4fe4 Fix singletons - only go through orte_proc_info_init once! Session dir is calling it be sure it is filled in, which was causing us to reset the fields.
This commit was SVN r17944.
2008-03-25 02:10:42 +00:00
Ralph Castain
dc7f45dafd Remove the obsolete and largely unused orte_system_info structure. The only fields that were used in that struct were nodeid and nodename - these have been transferred to the orte_process_info structure.
Only one place used the user name field - session_dir, when formulating the name of the top-level directory. Accordingly, the code for getting the user's id has been moved to the session_dir code.

This commit was SVN r17926.
2008-03-23 23:10:15 +00:00
Ralph Castain
ff99aa054f In order to prevent orphaned processes when using non-unity routing methods, the procs need to realize that their local daemon is a critical connection - if that connection unexpectedly closes, they need to terminate.
This commit adds definition for a "lifeline" connection. For an HNP, there is no lifeline, so the lifeline proc is NULL. For a daemon, the lifeline is the HNP - the daemon should abort if it loses that connection.

For a proc using unity routed, the lifeline is the HNP since it connects directly to the HNP.

For a proc using tree routed, the lifeline is the local daemon.

Adjusted OOB to call abort if the lifeline (as opposed to HNP) connection is lost.

This commit was SVN r17761.
2008-03-06 15:30:44 +00:00
Ralph Castain
d70e2e8c2b Merge the ORTE devel branch into the main trunk. Details of what this means will be circulated separately.
Remains to be tested to ensure everything came over cleanly, so please continue to withhold commits a little longer

This commit was SVN r17632.
2008-02-28 01:57:57 +00:00
Ralph Castain
3dbd4d9be7 Squeeeeeeze the launch message. This is the message sent to the daemons that provides all the data required for launching their local procs. In reorganizing the ODLS framework, I discovered that we were sending a significant amount of unnecessary and repeated data. This commit resolves this by:
1. taking advantage of the fact that we no longer create the launch  message via a GPR trigger. In earlier times, we had the GPR create the launch message based on a subscription. In that mode of operation, we could not guarantee the order in which the data was stored in the message - hence, we had no choice but to parse the message in a loop that checked each value against a list of possible "keys" until the corresponding value was found.

Now, however, we construct the message "by hand", so we know precisely what data is in each location in the message. Thus, we no longer need to send the character string "keys" for each data value any more. This represents a rather large savings in the message size - to give you an example, we typically would use a 30-char "key" for a 2-byte data value. As you can see, the overhead can become very large.

2. sending node-specific data only once. Again, because we used to construct the message via subscriptions that were done on a per-proc basis, the data for each node (e.g., the daemon's name, whether or not the node was oversubscribed) would be included in the data for each proc. Thus, the node-specific data was repeated for every proc.

Now that we construct the message "by hand", there is no reason to do this any more. Instead, we can insert the data for a specific node only once, and then provide the per-proc data for that node. We therefore not only save all that extra data in the message, but we also only need to parse the per-node data once.

The savings become significant at scale. Here is a comparison between the revised trunk and the trunk prior to this commit (all data was taken on odin, using openib, 64 nodes, unity message routing, tested with application consisting of mpi_init/mpi_barrier/mpi_finalize, all execution times given in seconds, all launch message sizes in bytes):

Per-node scaling, taken at 1ppn:

#nodes           original trunk                         revised trunk
             time               size                time               size
      1      0.10                819                0.09                564
      2      0.14               1070                0.14                677
      3      0.15               1321                0.14                790
      4      0.15               1572                0.15                903
      8      0.17               2576                0.20               1355
     16      0.25               4584                0.21               2259
     32      0.28               8600                0.27               4067
     64      0.50              16632                0.39               7683

Per-proc scaling, taken at 64 nodes

   ppn             original trunk                         revised trunk
              time               size                time               size
      1       0.50              16669                0.40               7720
      2       0.55              32733                0.54              11048
      3       0.87              48797                0.81              14376
      4       1.0               64861                0.85              17704


Condensing those numbers, it appears we gained:

per-node message size: 251 bytes/node -> 113 bytes/node

per-proc message size: 251 bytes/proc  -> 52 bytes/proc

per-job message size:  568 bytes/job -> 399 bytes/job 
(job-specific data such as jobid, override oversubscribe flag, total #procs in job, total slots allocated)

The fact that the two pre-commit trunk numbers are the same confirms the fact that each proc was containing the node data as well. It isn't quite the 10x message reduction I had hoped to get, but it is significant and gives much better scaling.

Note that the timing info was, as usual, pretty chaotic - the numbers cited here were typical across several runs taken after the initial one to avoid NFS file positioning influences.

Also note that this commit removes the orte_process_info.vpid_start field and the handful of places that passed that useless value. By definition, all jobs start at vpid=0, so all we were doing is passing "0" around. In fact, many places simply hardwired it to "0" anyway rather than deal with it.

This commit was SVN r16428.
2007-10-11 15:57:26 +00:00
Ralph Castain
54b2cf747e These changes were mostly captured in a prior RFC (except for #2 below) and are aimed specifically at improving startup performance and setting up the remaining modifications described in that RFC.
The commit has been tested for C/R and Cray operations, and on Odin (SLURM, rsh) and RoadRunner (TM). I tried to update all environments, but obviously could not test them. I know that Windows needs some work, and have highlighted what is know to be needed in the odls process component.

This represents a lot of work by Brian, Tim P, Josh, and myself, with much advice from Jeff and others. For posterity, I have appended a copy of the email describing the work that was done:

As we have repeatedly noted, the modex operation in MPI_Init is the single greatest consumer of time during startup. To-date, we have executed that operation as an ORTE stage gate that held the process until a startup message containing all required modex (and OOB contact info - see #3 below) info could be sent to it. Each process would send its data to the HNP's registry, which assembled and sent the message when all processes had reported in.

In addition, ORTE had taken responsibility for monitoring process status as it progressed through a series of "stage gates". The process reported its status at each gate, and ORTE would then send a "release" message once all procs had reported in.

The incoming changes revamp these procedures in three ways:

1. eliminating the ORTE stage gate system and cleanly delineating responsibility between the OMPI and ORTE layers for MPI init/finalize. The modex stage gate (STG1) has been replaced by a collective operation in the modex itself that performs an allgather on the required modex info. The allgather is implemented using the orte_grpcomm framework since the BTL's are not active at that point. At the moment, the grpcomm framework only has a "basic" component analogous to OMPI's "basic" coll framework - I would recommend that the MPI team create additional, more advanced components to improve performance of this step.

The other stage gates have been replaced by orte_grpcomm barrier functions. We tried to use MPI barriers instead (since the BTL's are active at that point), but - as we discussed on the telecon - these are not currently true barriers so the job would hang when we fell through while messages were still in process. Note that the grpcomm barrier doesn't actually resolve that problem, but Brian has pointed out that we are unlikely to ever see it violated. Again, you might want to spend a little time on an advanced barrier algorithm as the one in "basic" is very simplistic.

Summarizing this change: ORTE no longer tracks process state nor has direct responsibility for synchronizing jobs. This is now done via collective operations within the MPI layer, albeit using ORTE collective communication services. I -strongly- urge the MPI team to implement advanced collective algorithms to improve the performance of this critical procedure.


2. reducing the volume of data exchanged during modex. Data in the modex consisted of the process name, the name of the node where that process is located (expressed as a string), plus a string representation of all contact info. The nodename was required in order for the modex to determine if the process was local or not - in addition, some people like to have it to print pretty error messages when a connection failed.

The size of this data has been reduced in three ways:

(a) reducing the size of the process name itself. The process name consisted of two 32-bit fields for the jobid and vpid. This is far larger than any current system, or system likely to exist in the near future, can support. Accordingly, the default size of these fields has been reduced to 16-bits, which means you can have 32k procs in each of 32k jobs. Since the daemons must have a vpid, and we require one daemon/node, this also restricts the default configuration to 32k nodes.

To support any future "mega-clusters", a configuration option --enable-jumbo-apps has been added. This option increases the jobid and vpid field sizes to 32-bits. Someday, if necessary, someone can add yet another option to increase them to 64-bits, I suppose.

(b) replacing the string nodename with an integer nodeid. Since we have one daemon/node, the nodeid corresponds to the local daemon's vpid. This replaces an often lengthy string with only 2 (or at most 4) bytes, a substantial reduction.

(c) when the mca param requesting that nodenames be sent to support pretty error messages, a second mca param is now used to request FQDN - otherwise, the domain name is stripped (by default) from the message to save space. If someone wants to combine those into a single param somehow (perhaps with an argument?), they are welcome to do so - I didn't want to alter what people are already using.

While these may seem like small savings, they actually amount to a significant impact when aggregated across the entire modex operation. Since every proc must receive the modex data regardless of the collective used to send it, just reducing the size of the process name removes nearly 400MBytes of communication from a 32k proc job (admittedly, much of this comm may occur in parallel). So it does add up pretty quickly.


3. routing RML messages to reduce connections. The default messaging system remains point-to-point - i.e., each proc opens a socket to every proc it communicates with and sends its messages directly. A new option uses the orteds as routers - i.e., each proc only opens a single socket to its local orted. All messages are sent from the proc to the orted, which forwards the message to the orted on the node where the intended recipient proc is located - that orted then forwards the message to its local proc (the recipient). This greatly reduces the connection storm we have encountered during startup.

It also has the benefit of removing the sharing of every proc's OOB contact with every other proc. The orted routing tables are populated during launch since every orted gets a map of where every proc is being placed. Each proc, therefore, only needs to know the contact info for its local daemon, which is passed in via the environment when the proc is fork/exec'd by the daemon. This alone removes ~50 bytes/process of communication that was in the current STG1 startup message - so for our 32k proc job, this saves us roughly 32k*50 = 1.6MBytes sent to 32k procs = 51GBytes of messaging.

Note that you can use the new routing method by specifying -mca routed tree - if you so desire. This mode will become the default at some point in the future.


There are a few minor additional changes in the commit that I'll just note in passing:

* propagation of command line mca params to the orteds - fixes ticket #1073. See note there for details.

* requiring of "finalize" prior to "exit" for MPI procs - fixes ticket #1144. See note there for details.

* cleanup of some stale header files

This commit was SVN r16364.
2007-10-05 19:48:23 +00:00
Tim Prins
c46ed1d5d4 Make it so the universe size is passed through the ODLS instead of through a gpr trigger during MPI init. This matches what is currently being done with the app number.
The default odls has been updated and works fine. The process odls has been updated, but I could not verify its operation. The bproc ODLS has not been updated yet. Ralph will look at it soon.

This commit was SVN r15257.
2007-07-02 01:33:35 +00:00
Ralph Castain
d9acc93efa Compute and pass the local_rank and local number of procs (in that proc's job) on the node.
To be precise, given this hypothetical launching pattern:

host1: vpids 0, 2, 4, 6
host2: vpids 1, 3, 5, 7

The local_rank for these procs would be:

host1: vpids 0->local_rank 0, v2->lr1, v4->lr2, v6->lr3
host2: vpids 1->local_rank 0, v3->lr1, v5->lr2, v7->lr3

and the number of local procs on each node would be four. If vpid=0 then does a comm_spawn of one process on host1, the values of the parent job would remain unchanged. The local_rank of the child process would be 0 and its num_local_procs would be 1 since it is in a separate jobid.

I have verified this functionality for the rsh case - need to verify that slurm and other cases also get the right values. Some consolidation of common code is probably going to occur in the SDS components to make this simpler and more maintainable in the future.

This commit was SVN r14706.
2007-05-21 14:30:10 +00:00