ompi
ompi copied to clipboard
open-mpi
Update OpenPMIx and PRRTe submodules.
OpenPMIx commits:
c8bf473b - Add action to check commits on pull requests 03d9d092 - llvm/oneapi: fixes to bring pmix up to iso c99 66404931 - Fix the check_cli_option code 74bba6c3 - Update autogen.pl to check for submodules f45410f9 - build: Import and use external Autoconf repo 5e7ab95b - Update Github actions to use submodules bd679b02 - Enhance the performance of the var_scope_push/pop script cfe12e8b - pnet/nvd: Fix macro escaping issue 7aed4759 - Export the output_stream_t class declaration 1662c105 - Add improved debug and correct param passing to pmix_init_util e66c29b9 - Mark that proc arrays being passed have been sorted 0f8ef592 - Remove stale m4 and unimplemented function declaration db800a63 - Ensure tools relay events to their server 073251a4 - Add a couple more definitions 13377c57 - Fix Coverity CID #378400 11925eee - Clean up left over .gitignore entry after mca/common/* removal in https://github.com/openpmix/openpmix/commit/4f8a4bdc 6927bc40 - Convert the MCA parameter "pmix_mca_base_component_show_load_errors" to be a flexible mechanism to specify when (and when not) to emit warnings about errors when trying to load DSO components. 8465992f - Consistently use PMIx_Error_string in client example
PRRTe commits:
02ec0841 - schizo/ompi: Remove --display-devel-allocation 66147afe - schizo/ompi: Fix --display topo 1e5ab45c - ompi/schizo: Fix --do-not-launch and --show-progress. c277025b - Update the schizo/ompi component 504c97736e - Fix setting of runtime options c0802f4c34 - Update autogen.pl to check for submodules de3dfab647 - build: Import and use external Autoconf repo 70ec5fed86 - Fix a few minor things bed6eb42d3 - Cleanup code clause identification 3406d5fadd - Replace prte_output with pmix_output 322abf423a - Remove the --with-ft configure option f0900fd906 - Rename component libraries and component struct names 1a108ab6a7 - Handle default MCA params 30bbe0155b - Remove stale m4 898ecef552 - Delete PRRTE MCA infrastructure e7cb6e1361 - Remove leftover BINDTONONE SET debug code 986ffe396d - Update to use PMIx CLI option macro and "check true" 51bd9c3b50 - Add/modify some spawn-related tests e8640f1fcc - Revise passing of runtime options adc8fd769d - Fix hetero topology operations 1a90bcd32e - Deprecated the --stream-buffering CLI option 1207c6494a - stream-buffering is an OMPI option, so make it an MCA
Signed-off-by: Austen Lauria [email protected]
![azure-pipelines[bot] avatar](https://avatars.githubusercontent.com/in/9426?v=4 "Published by a pub.dev verified publisher"){kind=small}
Just an FYI: there is no --with-ft configure option for PRRTE any more in the master branch. I have likewise removed the components that depended on that setting. I'm working now on the FT support - IIRC, you should be getting failure notifications currently, but you do have to set the runtime options to do so (and I don't think the OMPI schizo component does that yet). I'm planning to update soon and include the required fixes to schizo/ompi. Will keep you updated.
Hmmm...not sure why you'd be segfaulting here. Let me try that cmd line locally - could be there is a problem with your translation code.
You'll need https://github.com/openpmix/prrte/pull/1511 to fix it - missing a simple update in the schizo/ompi component.
Updated PRRTE submodule pointer to pickup 4 more commits:
02ec0841 - schizo/ompi: Remove --display-devel-allocation
66147afe - schizo/ompi: Fix --display topo
1e5ab45c - ompi/schizo: Fix --do-not-launch and --show-progress.
c277025b - Update the schizo/ompi component
@awlauria I would suggest updating both submodules to the new HEAD of their respective master branches as they contain some commits from @jsquyres to support some PRs over here. If you could then get this committed, I would appreciate it so I can see the MTT results before releasing new PMIx/PRRTE versions.
@janjust @artemry-nv I think that the Mellanox CI test failed because it didn't git clone --recursive to get the Open MPI repo. This is especially important now that PMIx and PRTE have submodules themselves (I think this PR is updating Open MPI's PMIx's and PRTE's submodules to the point where they now have submodules, which is probably why this is the first time we're seeing this issue).
(carried this comment over from https://github.com/open-mpi/ompi/pull/10834#issuecomment-1256290064 -- that PR was closed and replaced with this one)
@janjust @artemry-nv I think that the Mellanox CI test failed because it didn't git clone --recursive to get the Open MPI repo. This is especially important now that PMIx and PRTE have submodules themselves (I think this PR is updating Open MPI's PMIx's and PRTE's submodules to the point where they now have submodules, which is probably why this is the first time we're seeing this issue).
(carried this comment over from #10834 (comment) -- that PR was closed and replaced with this one)
As far as I see submodules are already enabled https://github.com/open-mpi/ompi/blob/afd23c97d07e4378eafbc7cae5ff69ee17c52917/.ci/mellanox/azure-pipelines.yml#L28
Error message:
2. Checking for git submodules
=== Submodule: config/oac
==> ERROR: Missing
The submodule "config/oac" is missing.
Is everything ok with submodules? Is it a new one?
@artemry-nv config/oac is a new submodule; just added within the last week or two.
@artemry-nv
config/oacis a new submodule; just added within the last week or two.
Looks strange, as far as I see it works fine on main branch. oac submodule haven't been updated within this PR also. Could you please elaborate how you've updated git submodules which are in scope of this PR (exact git commands)?
CC @awlauria
I see, now there're two config/oac submodules - one is in root of ompi and another inside prrte.
Need to change https://github.com/open-mpi/ompi/blob/31953c1ba425dd1fd4016d76a80fd6b88720caaf/.ci/mellanox/azure-pipelines.yml#L28
to
submodules: recursive
according to https://learn.microsoft.com/en-us/azure/devops/pipelines/yaml-schema/steps-checkout?view=azure-pipelines