Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Move the registration of MCA params out of the init of the var system… #1650

Merged
merged 1 commit into from
May 20, 2016
Merged

Move the registration of MCA params out of the init of the var system… #1650

merged 1 commit into from
May 20, 2016

Conversation

rhc54
Copy link
Contributor

@rhc54 rhc54 commented May 7, 2016

… - put them in with the rest of the OPAL MCA param registrations

@rhc54
Copy link
Contributor Author

rhc54 commented May 7, 2016

@jjhursey @jladd-mlnx

I'm going to need some assistance in tracking this problem down. I'm not sure what this test is supposed to do, but I'm guessing it has some relation to reading and processing of aggregate MCA param files? I've tried tracing that code, but there are few comments about what is supposed to happen where, and it traverses several locations - so I'm having no luck figuring it out.

Basically, I think the problem is that we are registering an MCA param too early. This PR moves the two most obvious cases to a better place, but that still didn't fix this problem.

I'd appreciate it if someone familiar with the AMCA (and I guess we now have an env file?) could take a look and suggest a fix.

@rhc54
Copy link
Contributor Author

rhc54 commented May 15, 2016

Adding @hjelmn since he also worked on the param code.

I surrender - the code for reading MCA param files loops over and around and thru itself across multiple files. The -am option itself is rather confusing as it claims to set a path, but seems to specify the file within a path?? Very confusing.

I'm afraid this is going to remain dead until someone who remembers something about how all this is supposed to work can attack it. When they do, please (a) untangle the spaghetti, and (b) comment the code so someone else can figure out what is going on.

Meantime, please disable the AMCA-related tests.

@jjhursey
Copy link
Member

Related PR #1675

… - put them in with the rest of the OPAL MCA param registrations

Take another shot at untangling the spaghetti

orterun: fix for command line parsing

orte-submit calls opal_init_util () before parsing out MCA command line
options (-mca, -am, etc). This prevents mpirun from setting opal MCA
variables for some frameworks as well as the MCA base. This is because
when a framework is opened all of its variables are set to read-only.
Eventually we want to lift this restriction on some MCA variables but
since -mca is affected we must parse out the MCA command line options
before opal_init_util(). This commit fixes the bug by adding a new
option to opal_cmd_line_parse (ignore unknown option) so orte-submit
can pre-parse the command line for MCA options.

Signed-off-by: Nathan Hjelm <[email protected]>

Minor cleanups to avoid releasing/recreating the cmd line
@rhc54
Copy link
Contributor Author

rhc54 commented May 20, 2016

This doesn't fully solve the envlist problem, but these are all necessary steps, so let's commit that part.

@rhc54 rhc54 merged commit 5ba8344 into open-mpi:master May 20, 2016
@rhc54 rhc54 deleted the topic/envlist branch May 20, 2016 17:00
@lanl-ompi
Copy link
Contributor

Test FAILed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants