search

Exawind / nalu-wind

Solver for wind farm simulations targeting exascale computational platforms
https://nalu-wind.readthedocs.io
Other
124 stars 85 forks source link

probe name #677

Open petesilk opened 4 years ago

petesilk commented 4 years ago

I am running a case with multiple line_of_site_specifications: probes. They have names such as "- name: probeM0", "- name: probeM2", etc. However, if I include the name "- name: probeM1", the case will not start. Does a core dump and creates the log and error files attached. If I change that one probe name, even simply by removing the "o" to make it "prbeM1", the case runs fine.

is "probeM1" a reserved or special name? why does this not work?

Thank you. user_script.stderr.txt TM.log

petesilk commented 4 years ago

Note that if I start from a mesh, instead of trying to use the "restart" from a previous solution.....I do not have the above problem.

psakievich commented 4 years ago

Interesting. when you remove the "o" are you getting probe data from your probes? Also did you have these probes with the run you used to create your restart dataset?

petesilk commented 4 years ago

Yes, when I remove the “o” I do get data.

No, did not have these particular probes on in the run that created the restart data. That said, other new probes (that were not on in the run that generated the restart file) with names other than probeM1 did work fine.

Thank you, Peter

From: psakievich notifications@github.com Sent: Thursday, August 13, 2020 5:03 PM To: Exawind/nalu-wind nalu-wind@noreply.github.com Cc: Silkowski, Peter (GE Research, US) silkowski@ge.com; Author author@noreply.github.com Subject: EXT: Re: [Exawind/nalu-wind] probe name (#677)

Interesting. when you remove the "o" are you getting probe data from your probes? Also did you have these probes with the run you used to create your restart dataset?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/Exawind/nalu-wind/issues/677#issuecomment-673709097, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AQJAUQD2IXQFRJ5FKDZRVGLSARIJRANCNFSM4P6SHCDA.

spdomin commented 4 years ago

I was just looking at Nalu-Wind and saw this bug report. The initial data probes implementation was really sensitive to restarting. There were plenty of use-cases that worked, however, may common ones that failed. You may want to check out Nalu for the fix. The nuance is the mapping of the probe to a mpi rank and and allowing for modifications at restart in addition to more details. The Nalu commit history has more info.

https://github.com/NaluCFD/Nalu/commit/0a57ef5389c45b1a8eb8949abe327347993b7416

petesilk commented 4 years ago

Thanks much.

From: Stefan P. Domino notifications@github.com Sent: Friday, September 04, 2020 4:08 PM To: Exawind/nalu-wind nalu-wind@noreply.github.com Cc: Silkowski, Peter (GE Research, US) silkowski@ge.com; Author author@noreply.github.com Subject: EXT: Re: [Exawind/nalu-wind] probe name (#677)

I was just looking at Nalu-Wind and saw this bug report. The initial data probes implementation was really sensitive to restarting. There were plenty of use-cases that worked, however, may common ones that failed. You may want to check out Nalu for the fix. The nuance is the mapping of the probe to a mpi rank and and allowing for modifications at restart in addition to more details. The Nalu commit history has more info.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/Exawind/nalu-wind/issues/677#issuecomment-687356875, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AQJAUQGT4V2XQHWA6SVLNW3SEFCLHANCNFSM4P6SHCDA.

psakievich commented 3 weeks ago

@rcknaus and @pscrozi it looks like we have not applied the fix that @spdomin suggested. It would be worth updating the code for more parity with NaluCFD. However, it seems there are also workaround for the users.