1
1
Граф коммитов

189 Коммитов

Автор SHA1 Сообщение Дата
Ralph Castain
368684bd63 Revert e9bc293 and try a different approach for scalably dealing with hetero clusters. Have each orted send back its topo "signature". If mpirun detects that this signature has not been seen before, then ask for that daemon to send back its full topology description. This allows the system to only get the topology once for each unique topo in the cluster.
Cleanup a typo, and remove no longer needed MCA params for hetero nodes and hetero apps. Hetero nodes will always be automatically detected. We don't support a mix of 32 and 64 bit apps

Modify the orte_node_t to use orte_topology_t instead of hwloc_topology_t, updating all the places that use it. Ensure that we properly update topology when we see a different one on a compute node.

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2017-01-18 10:22:15 -08:00
Nathan Hjelm
110840fc87 ess/hnp: add support for forwarding additional signals (#2712)
* ess/hnp: add support for forwarding additional signals

This commit adds support to the hnp ess module to forward additional
signals beyond the default SIGUSR1, SIGUSR2, SIGSTP, and SIGCONT.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>

* Generalize this a bit to allow a broader range of signals to be forwarded. Turns out that SIGURG is now a "standard" signal, though the value differs across systems. So setup to forward it (and some friends) if they are defined. Allow users to provide the signal name (instead of the integer value) as the value of even the more common signals does vary across systems. Don't limit the number that can be supported.

Signed-off-by: Ralph Castain <rhc@open-mpi.org>

* ess/hnp: fix some bugs in the signal forwarding code

This commit fixes two bugs:

 - signals_set needs to be set even if no signals are being
   forwarded. If it is not set we will SEGV in libevent if
   ess_hnp_forward_signals == none.

 - SIGTERM and SIGHUP are handled with a different type of handler. Do
   not allow the user to specify these to be forwarded.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>

* We are sure to get "dinged" if error messages aren't nicely output via show_help, so do so here

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2017-01-12 10:09:41 -07:00
Ralph Castain
93e4935902 Be a tad more cautious before releasing objects when running in DVM mode
Fixes #2700

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2017-01-10 14:04:27 -08:00
Gilles Gouaillardet
58f2a764f9 ess/hnp: plug memory leaks
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
2017-01-06 11:35:59 +09:00
Ralph Castain
b8c5d1ad88 Update the routed components as we no longer need to init_routes. Fixes case of direct launch via srun
Signed-off-by: Ralph Castain <rhc@open-mpi.org>

Signed-off-by: Ralph Castain <rhc@open-mpi.org>
2016-10-31 12:38:13 -07:00
Ralph Castain
649301a3a2 Revise the routed framework to be multi-select so it can support the new conduit system. Update all calls to rml.send* to the new syntax. Define an orte_mgmt_conduit for admin and IOF messages, and an orte_coll_conduit for all collective operations (e.g., xcast, modex, and barrier).
Still not completely done as we need a better way of tracking the routed module being used down in the OOB - e.g., when a peer drops connection, we want to remove that route from all conduits that (a) use the OOB and (b) are routed, but we don't want to remove it from an OFI conduit.
2016-10-23 21:52:39 -07:00
Ralph Castain
a2919174d0 Bring the RML modifications across. This is the first step in a revamp of the ORTE messaging subsystem to support fabric-based communications during launch and wireup phases. When completed, the grpcomm and plm frameworks will each have their own "conduit" for communication - each conduit corresponds to a particular RML messaging transport. This can be the active OOB-based component, or a provider from within the RML/OFI component. Messages sent down the conduit will flow across the associated transport.
Multiple conduits can exist at the same time, and can even point to the same base transport. Each conduit can have its own characteristics (e.g., flow control) based on the info keys provided to the "open_conduit" call. For ease during the transition period, the "legacy" RML interfaces remain as wrappers over the new conduit-based APIs using a default conduit opened during orte_init - this default conduit is tied to the OOB framework so that current behaviors are preserved. Once the transition has been completed, a one-time cleanup will be done to update all RML calls to the new APIs and the "legacy" interfaces will be deleted.

While we are at it: Remove oob/usock component to eliminate the TMPDIR length problem - get all working, including oob_stress
2016-10-11 16:01:02 -07:00
Artem Polyakov
81195ab724 Several fixes related to session directories:
* enable OMPI to retrieve paths from RM through PMIx
* cleanups related to tempdirs.
2016-09-05 07:48:44 +03:00
Ralph Castain
ee56d9dc1a Shorten the session directory name as some OS's are now providing unusually long temp directory names, causing us to overflow the sockaddr field 2016-07-05 14:59:50 -07:00
Ralph Castain
6ac7929bd0 Extend the schizo framework to allow definition of CLI options by environment. Refactor orterun to mesh with the orted_submit code, thus improving code reuse. Eliminate the orte-submit tool as orterun can now meet that need.
Cleanups per @jjhursey review
2016-05-01 11:30:25 -07:00
Ralph Castain
2432daf065 Some minor cleanups of a memory leak and error output 2016-04-08 07:46:18 -07:00
Rainer Keller
52080a5736 As per the pull request to pmix/master:
https://github.com/pmix/master/pull/71

Have OMPI's current version of pmix120 nicely fail in case of
too long sun_path (longer than 108 or in case of OSX 103 chars).
And have OMPI return proper error messages with hints how to
amend.
2016-04-07 22:12:53 +02:00
rhc54
6756e19aa2 Merge pull request #1457 from anandhis/master
rml changes
2016-03-24 15:17:29 -07:00
Ralph Castain
a4c8e8c28a Cleanup the proposed change:
* qos framework is moving to the scon layer and is no longer required in ORTE

* remove the rml/ftrm component as we now have multiple active components, and so the wrapper needs to be rethought

* no need for separating the "base" from "API" module definition. The two are identical

* move the "stub" functions into their own file for cleanliness

* general cleanup to meet coding standards

* cleanup some logic in the stubs
2016-03-10 13:14:17 -08:00
Ralph Castain
d72c1c72ff Do not push child processes into separate process groups so that any host RM can still "see" them, and ensure that any signal sent to the orted's themselves will be provided to all child processes. Forward all signals from mpirun to the child processes, removing the old MCA parameter required to turn that behavior "on". 2016-03-06 17:55:09 -08:00
Ralph Castain
4a55fba414 Fix registration of error handlers thru the pmix120 component. A thread-shift operation was hanging on the sync_event_base, which made it dependent on someone calling opal_progress. Unfortunately, a process in "sleep" or spinning outside the MPI library won't do that, and so we never complete errhandler registration. 2016-03-02 15:01:01 -08:00
Ralph Castain
011403c04a Fix a number of issues, some of which have lingered for a long time:
* provide a more reliable way of determining that a process is a singleton by leveraging the schizo framework. Add new components for slurm, alps, and orte to detect when we are in a managed environment, and if we have been launched by mpirun or a native launcher. Set the correct envars to control ess and pmix selection in each case.

* change the relative priority of the pmix120 and pmix112 components to make pmix120 the default

* fix singleton comm-spawn by correctly setting the num_apps field of the orte_job_t created by the daemon - this fixes a segfault in register_nspace on newly created daemons

* ensure orterun doesn't propagate any ess or pmix directives in its environment

* Cleanup a few valgrind issues and memory leaks

* Fix a race condition that prevented the client from completing notification registrations (missing thread shift)

* Ensure the shizo/alps component detects launch by mpirun
2016-03-01 06:53:00 -08:00
Ralph Castain
d653cf2847 Convert the orte_job_data pointer array to a hash table so it doesn't grow forever as we run lots and lots of jobs in the persistent DVM. 2016-02-21 11:55:49 -08:00
Ralph Castain
94ffe10808 Do not override any external settings for PMIx component selection 2015-12-21 08:36:12 -08:00
Ralph Castain
0140ff048d Now that we have an "isolated" PLM component, we cannot just let rsh silently decline to run when it cannot find a launch agent - if we do, then we will -always- run on the local node. So if the user specifies a launch agent and we can't find it, then generate a pretty error message, report a fatal error back to the component select, and exit out.
This required modifying the mca_component_select function to actually check the return code on a component query - it was blissfully ignoring it.

Also do a little cleanup to avoid bombarding the user with multiple error messages.

Thanks to Patrick Begou for reporting the problem
2015-09-24 07:16:48 -07:00
Ralph Castain
f872e99315 Fix orte-submit so it allows application procs to select the correct ess component. Protect orte_data_server from multiple calls to finalize. 2015-09-21 20:31:57 -07:00
Ralph Castain
1cdb86b8c7 Cleanup s1 and s2 components, and ensure that mpirun and orteds only use non-direct-launch pmix components. 2015-09-08 18:37:09 -07:00
rhc54
665b30376a Merge pull request #868 from rhc54/topic/hwloc
Remove OPAL_HAVE_HWLOC qualifier and error out if --without-hwloc is given
2015-09-04 17:58:07 -07:00
Ralph Castain
d97bc29102 Remove OPAL_HAVE_HWLOC qualifier and error out if --without-hwloc is given 2015-09-04 16:54:40 -07:00
Ralph Castain
f6948c2bb4 Sync with PMIx master 43e45c3. Get multi-node publish/lookup/unpublish working 2015-09-04 10:07:17 -07:00
Ralph Castain
cf6137b530 Integrate PMIx 1.0 with OMPI.
Bring Slurm PMI-1 component online
Bring the s2 component online

Little cleanup - let the various PMIx modules set the process name during init, and then just raise it up to the ORTE level. Required as the different PMI environments all pass the jobid in different ways.

Bring the OMPI pubsub/pmi component online

Get comm_spawn working again

Ensure we always provide a cpuset, even if it is NULL

pmix/cray: adjust cray pmix component for pmix

Make changes so cray pmix can work within the integrated
ompi/pmix framework.

Bring singletons back online. Implement the comm_spawn operation using pmix - not tested yet

Cleanup comm_spawn - procs now starting, error in connect_accept

Complete integration
2015-08-29 16:04:10 -07:00
Ralph Castain
0b1d4b62be Cleanup some cruft and update to coordinate with CM operations:
* don't pass --tree-spawn to the orted cmd line. If someone doesn't want tree-spawn, it shows up as an MCA param anyway
* ensure state/orted component disqualifies itself from CM operations
* clarify the DVM proc_type definitions
* ensure we stop littering the tmp dir with session directories
2015-08-12 10:32:14 -07:00
Ralph Castain
869041f770 Purge whitespace from the repo 2015-06-23 20:59:57 -07:00
Ralph Castain
1f8de276de Consolidate all the QOS changes into one clean commit 2015-05-06 19:48:42 -07:00
Nathan Hjelm
45e053dbce orte: use C99 subobject naming for component initialization
This commit helps future-proof orte components by initializing each
component member by name.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
2015-04-18 10:29:58 -06:00
Ralph Castain
cd686057f6 If the HNP is on a coprocessor, record it so we don't get an error log later 2015-04-11 15:30:15 -07:00
Ralph Castain
43a3baad5e Ensure we use the first compute node's topology for mapping
Don't filter the topology by cpuset if you are mpirun until you know that no other compute nodes are involved. This deals with the corner case where mpirun is executing on a node of different topology from the compute nodes.

Simplify - don't mandate that all cpus in the given cpuset be present on every node. We can then run everything thru the filter as before, which ensures that any procs run on mpirun are also contained within the specified cpuset.

Correctly count the number of available PUs under each object when given a cpuset

Fix the default binding settings, and correctly count PUs when no cpuset is given

Ensure the binding policy gets set in all cases
2015-03-19 16:30:36 -07:00
Howard Pritchard
bf89131f9e add owner files to opa/ompi/orte mca directories
This commit adds an owner file in each of the component directories
for each framework.  This allows for a simple script to parse
the contents of the files and generate, among other things, tables
to be used on the project's wiki page.  Currently there are two
"fields" in the file, an owner and a status.  A tool to parse
the files and generate tables for the wiki page will be added
in a subsequent commit.
2015-02-22 15:10:23 -07:00
Ralph Castain
028b00154d Complete implementation of the schizo framework to support OMPI component 2015-01-27 09:29:42 -06:00
Ralph Castain
bb529ebd8e Revise the way we handle hetero nodes as users are finding this (a) a significant surprise, and (b) confusing as to when it is required. So try to automate it a bit by creating a topology "signature" that mpirun can share on the cmd line with the remote daemons, thus allowing them to check to see if they match. This isn't comprehensive of course - for now, it only checks the number of each type of hwloc object on the node. This is good enough to pickup major differences (e.g., where we have different numbers of sockets or assigned core bindings).
Retain the hetero-nodes flag for those cases where the user *knows* that there are differences and our automated system isn't good enough to see it.

Will obviously require further refinement as we find out which variances it can detect, and which it cannot.
2014-12-08 15:38:14 -08:00
Ralph Castain
14cdb04327 Revise the ess/pmi selection logic as all APPs must select it, and no daemons. Cleanup some of the mca param levels in ess so we don't printout the topology quite as easily. 2014-12-01 21:19:11 -08:00
Ralph Castain
5cdbc00136 Re-enable the usock oob component. Ensure the TCP component promotes messages for other procs to the OOB base so that other components have a chance to send the relay. Seems to be passing MTT, so let's see how it works for others.
This commit was SVN r32650.
2014-08-30 19:33:46 +00:00
Ralph Castain
aec5cd08bd Per the PMIx RFC:
WHAT:    Merge the PMIx branch into the devel repo, creating a new
               OPAL “lmix” framework to abstract PMI support for all RTEs.
               Replace the ORTE daemon-level collectives with a new PMIx
               server and update the ORTE grpcomm framework to support
               server-to-server collectives

WHY:      We’ve had problems dealing with variations in PMI implementations,
               and need to extend the existing PMI definitions to meet exascale
               requirements.

WHEN:   Mon, Aug 25

WHERE:  https://github.com/rhc54/ompi-svn-mirror.git

Several community members have been working on a refactoring of the current PMI support within OMPI. Although the APIs are common, Slurm and Cray implement a different range of capabilities, and package them differently. For example, Cray provides an integrated PMI-1/2 library, while Slurm separates the two and requires the user to specify the one to be used at runtime. In addition, several bugs in the Slurm implementations have caused problems requiring extra coding.

All this has led to a slew of #if’s in the PMI code and bugs when the corner-case logic for one implementation accidentally traps the other. Extending this support to other implementations would have increased this complexity to an unacceptable level.

Accordingly, we have:

* created a new OPAL “pmix” framework to abstract the PMI support, with separate components for Cray, Slurm PMI-1, and Slurm PMI-2 implementations.

* Replaced the current ORTE grpcomm daemon-based collective operation with an integrated PMIx server, and updated the grpcomm APIs to provide more flexible, multi-algorithm support for collective operations. At this time, only the xcast and allgather operations are supported.

* Replaced the current global collective id with a signature based on the names of the participating procs. The allows an unlimited number of collectives to be executed by any group of processes, subject to the requirement that only one collective can be active at a time for a unique combination of procs. Note that a proc can be involved in any number of simultaneous collectives - it is the specific combination of procs that is subject to the constraint

* removed the prior OMPI/OPAL modex code

* added new macros for executing modex send/recv to simplify use of the new APIs. The send macros allow the caller to specify whether or not the BTL supports async modex operations - if so, then the non-blocking “fence” operation is used, if the active PMIx component supports it. Otherwise, the default is a full blocking modex exchange as we currently perform.

* retained the current flag that directs us to use a blocking fence operation, but only to retrieve data upon demand

This commit was SVN r32570.
2014-08-21 18:56:47 +00:00
Ralph Castain
daeb9b6c4f Some more cleanups. Remove direct references to ORTE by changing OMPI_CAST_ORTE_NAME -> OMPI_CAST_RTE_NAME. Ensure that ORTE tools (mpirun, orted, tools) set the OPAL proc structure fields so OPAL knows what is going on and uses the correct print functions (still need to fix the problem for non-MPI apps). Properly return uint32_t from the opal utilities instead of int32_t as that is what the ORTE process name fields contain.
Thanks to Gilles for pointing out some of the discrepancies.

This commit was SVN r32398.
2014-08-01 14:44:11 +00:00
Ralph Castain
8736a1c138 Per RFC:
http://www.open-mpi.org/community/lists/devel/2014/05/14822.php

Revamp the ORTE global data structures to reduce memory footprint and add new features. Add ability to control/set cpu frequency, though this can only be done if the sys admin has setup the system to support it (or you run as root).

This commit was SVN r31916.
2014-06-01 16:14:10 +00:00
Ralph Castain
11faab1091 The final step of the RFC: convert the <foo>libdir and friends to fit their respective code areas, and equate them all at the top. Note that we can't entirely separate things as the opal_install_dirs framework can't handle separated locations for the various trees.
This commit was SVN r31679.
2014-05-08 02:01:35 +00:00
Ralph Castain
87d809eefe Add a new "run-time controls" framework for setting controls on processes. Initially, just move the process binding code there under a new "hwloc" component. Additional components to support cgroups, power settings, etc. to follow
This commit was SVN r31633.
2014-05-05 19:22:06 +00:00
Ralph Castain
c4c9bc1573 As per the RFC:
http://www.open-mpi.org/community/lists/devel/2014/04/14496.php

Revamp the opal database framework, including renaming it to "dstore" to reflect that it isn't a "database". Move the "db" framework to ORTE for now, soon to move to ORCM

This commit was SVN r31557.
2014-04-29 21:49:23 +00:00
Jeff Squyres
d8715f1e3a Close 3 more fd's that were leaking into child processes.
Child processes now look clean; I can't find any more fd's that are
leaking from the parent to children.

Refs trac:4550

This commit was SVN r31515.

The following Trac tickets were found above:
  Ticket 4550 --> https://svn.open-mpi.org/trac/ompi/ticket/4550
2014-04-24 15:36:24 +00:00
Ralph Castain
8594f5d738 Correctly set a non-zero exit status when mpirun is terminated by signal
Fixes trac:4537

cmr=v1.8.1:reviewer=jsquyres

This commit was SVN r31423.

The following Trac tickets were found above:
  Ticket 4537 --> https://svn.open-mpi.org/trac/ompi/ticket/4537
2014-04-18 16:39:08 +00:00
Ralph Castain
8d72633acf Ensure that the session directory fields of orte_process_info have been initialized prior to cleaning up those directories as part of the initialization process that deals with stale session directory trees.
Fixes trac:4534

cmr=v1.8.1:reviewer=jsquyres

This commit was SVN r31421.

The following Trac tickets were found above:
  Ticket 4534 --> https://svn.open-mpi.org/trac/ompi/ticket/4534
2014-04-18 14:25:48 +00:00
Ralph Castain
a368e84e70 Per the RFC, remove the sensor framework from the ORTE code area, relocating it offsite to the ORCM code area. Also update some ignores to ensure we don't pickup crosstalk in components
This commit was SVN r31403.
2014-04-15 21:48:24 +00:00
Ralph Castain
f56f37d364 Shifting to an event-driven RTE raises some interesting issues during shutdown. We want the last messages to get thru, but also need to correctly shutdown the virtual machine. This requires a delicate balancing act across event priorities, and the need to check for termination conditions in places where related events get processed.
Change the priority of comm_failure and job_termination events to ensure we process final messages prior to terminating. Check for termination conditions when processing proc termination events as we may order proc termination when the daemon gets an exit command, but we can't see the proc actually terminate until we get out of that message event.

Jeff: probably easiest to review this by testing. I tested it under both Slurm and rsh on v1.7.5 as well as trunk

cmr=v1.7.5:reviewer=jsquyres:subject=resolve event priorities during VM shutdown

This commit was SVN r31042.
2014-03-12 16:49:58 +00:00
Ralph Castain
0ee38353ba In case there are stale session directories around, do a purge of the relevant session directory tree when an orted, HNP, or singleton start. This won't help in the case of direct-launched apps, but it's the best we can do.
cmr=v1.7.5:reviewer=jsquyres:subject=purge stale session dirs at startup

This commit was SVN r30642.
2014-02-09 02:10:31 +00:00
Adrian Reber
8c93ebffeb orte_snapc_base_select() wants to know if it is an application
The function

   int orte_snapc_base_select(bool seed, bool app);

wants to know if it called by an application or not. Therefore
it expects as second paremeter 'bool app'. It used to be
'!ORTE_PROC_IS_DAEMON' which is not always correct if it is
a tool or a HNP. This patch changes it to ORTE_PROC_IS_APP, which
has the correct information if it is an application.

This commit was SVN r30404.
2014-01-24 17:14:41 +00:00