opm-simulators icon indicating copy to clipboard operation
opm-simulators copied to clipboard

Add support for SLAVES keyword

Open hakonhagland opened this issue 1 year ago • 4 comments

Depends on upstream https://github.com/OPM/opm-common/pull/4114.

This is a first step to implement reservoir coupling

Currently the SLAVES keywords does nothing. The plan is to activate the keyword after GRUPMAST and GRUPSLAV has been implemented. This can hopefully be done using MPI_Comm_spawn() from the master, and then MPI communication between master and slaves.

hakonhagland avatar Jun 20 '24 21:06 hakonhagland

Since we SLAVES does not do anything at the moment, isn't it better to wait with merging this until we actually supports this feature. For testing you can always set the parsing-strictness=low

totto82 avatar Jun 24 '24 12:06 totto82

Since we SLAVES does not do anything at the moment, isn't it better to wait with merging this until we actually supports this feature.

@totto82 Good point.

hakonhagland avatar Jun 24 '24 12:06 hakonhagland

I convert this to draft until it is ready for merging

hakonhagland avatar Jun 24 '24 12:06 hakonhagland

Rebased

hakonhagland avatar Sep 09 '24 08:09 hakonhagland

@blattms I have removed the draft state of this PR, in accordance with https://github.com/OPM/opm-simulators/pull/5620#issuecomment-2368042583. Can you have a look again at this PR and see if we can get it merged?

hakonhagland avatar Sep 24 '24 06:09 hakonhagland

I would like to run jennkins here. Would be cool if you would rebase this onto current master. Without that some updated tests might actually fail.

blattms avatar Oct 02 '24 11:10 blattms

Just to clarify: This PR is there to simply mark the SLAVE keyword as supported?

IMHO, then this PR should actually be the last one in the sequence to support reservoir coupling. Once this is a user can run simulations with coupled reservoirs with the default option. He might expect that such a simulation then works and be surprised that it does not, because the feature is still work in progress.

I don't think that having the SLAVE keyword marked as supported is needed for testing. I would simply start flow with --parsing-strictness=low to make it continue the simulation with keywords marked as unsupported.

blattms avatar Oct 02 '24 11:10 blattms

@blattms this PR should actually be the last one in the sequence to support reservoir coupling.

Good point. I put this back into draft state then.

hakonhagland avatar Oct 02 '24 11:10 hakonhagland

Closing this since it will be replaced by #5643.

hakonhagland avatar Jan 21 '25 14:01 hakonhagland