V5.0.x bump submodules

[wenduwan]

Ingest prrte and pmix RC

[wenduwan]

This should address:

• PR 12101 broke support for PRTE_MCA_ras_base_launch_orted_on_hn #12150
• prrte submodule pointer bump broke --report-bindings #12143

[janjust]

@wenduwan

Example failed: 124
[1450](https://jenkins.open-mpi.org/jenkins/job/open-mpi.pull_request-v2/job/PR-12152/1/pipeline-console/?start-byte=803767&selected-node=108#log-1450)
+ echo 'Command was: timeout -s SIGSEGV 4m mpirun --get-stack-traces --timeout 180 --hostfile /home/ec2-user/workspace/pen-mpi.pull_request-v2_PR-12152/hostfile -np 2 --bind-to none  ./examples/hello_c'
[1451](https://jenkins.open-mpi.org/jenkins/job/open-mpi.pull_request-v2/job/PR-12152/1/pipeline-console/?start-byte=803767&selected-node=108#log-1451)
Command was: timeout -s SIGSEGV 4m mpirun --get-stack-traces --timeout 180 --hostfile /home/ec2-user/workspace/pen-mpi.pull_request-v2_PR-12152/hostfile -np 2 --bind-to none  ./examples/hello_c
[1452](https://jenkins.open-mpi.org/jenkins/job/open-mpi.pull_request-v2/job/PR-12152/1/pipeline-console/?start-byte=803767&selected-node=108#log-1452)
+ exit 124
[1453](https://jenkins.open-mpi.org/jenkins/job/open-mpi.pull_request-v2/job/PR-12152/1/pipeline-console/?start-byte=803767&selected-node=108#log-1453)
script returned exit code 124

Looks like we have a failure with the bumped pointers?

[bosilca]

We need to bump openpmix to ec54c21 to get the correct handling of -- in the arguments list (per openpmix/prrte#1886).

[rhc54]

My patch is actually on PRRTE

[wenduwan]

@rhc54 Thanks for the quick fix! Is this related to openpmix/prrte#1864?

Sorry if I missed the test case!

Now that we have a fix, do you plan to do another RC?

[rhc54]

Does anyone know why the CI actually failed? I don't see enough log output to know, and I certainly cannot reproduce a problem with running a simple "hello" program.

Is this related to openpmix/prrte#1864?

Sort of - but not really. The fix, at least, was unrelated to the fix for that specific issue.

Now that we have a fix, do you plan to do another RC?

I'd rather not do an RC for every individual problem/fix as that would become overly burdensome. I'd suggest just bumping to the head of the PMIx v4.2 and PRRTE v3.0 branches to pull in any fixes that relate to issues identified with OMPI's rc. We can then decide if another more formal round or rc's is needed.

[wenduwan]

Tracking prrte v3.0 branch in the latest revision. This will start the CI again. Will see will see.

[wenduwan]

Looking at CI failure.

[wenduwan]

The test appears flaky. I ran it twice, 1 passed 1 failed

(env) -bash-4.2$ ./run_example.sh
Hello, world, I am 0 of 2, (Open MPI v5.0.1a1, package: Open MPI [email protected] Distribution, ident: 5.0.1a1, repo rev: v5.0.0rc6-1154-gf1c35e8985, Unreleased developer copy, 183)
Hello, world, I am 1 of 2, (Open MPI v5.0.1a1, package: Open MPI [email protected] Distribution, ident: 5.0.1a1, repo rev: v5.0.0rc6-1154-gf1c35e8985, Unreleased developer copy, 183)
(env) -bash-4.2$ ./run_example.sh
+ export PATH=/home/ec2-user/ompi/install/bin:/home/ec2-user/env/bin:/opt/aws/neuron/bin:/opt/amazon/openmpi/bin/:/opt/amazon/efa/bin/:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/home/ec2-user/.local/bin:/home/ec2-user/bin
+ PATH=/home/ec2-user/ompi/install/bin:/home/ec2-user/env/bin:/opt/aws/neuron/bin:/opt/amazon/openmpi/bin/:/opt/amazon/efa/bin/:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/home/ec2-user/.local/bin:/home/ec2-user/bin
+ export LD_LIBRARY_PATH=/home/ec2-user/ompi/install/lib:
+ LD_LIBRARY_PATH=/home/ec2-user/ompi/install/lib:
+ timeout -s SIGSEGV 4m mpirun --get-stack-traces --timeout 180 --hostfile /home/ec2-user/PortaFiducia/hostfile -np 2 --bind-to none /home/ec2-user/ompi/examples/hello_c
(env) -bash-4.2$ echo $?
124

Checked v5.0.x branch. Pretty sure this is introduced with new prrte/pmix.

[rhc54]

Pretty sure this is introduced with new prrte/pmix.

Given the exact same behavior in #12156, which does not include a bump in submodule pointers, why would you conclude this?

[wenduwan]

No. I commented in the wrong thread.

The bump introduced a hang. Git bisect shows it's coming from this commit openpmix/prrte@5c19e5e

[rhc54]

Okay, bump it up again to pickup the fix.

[wenduwan]

@rhc54 Thank you!

I'm also testing it on my machine.

[wenduwan]

Local test passed!

[wenduwan]

@janjust CI passed after bumping the pointers.

[janjust]

@wenduwan good to merge?