|
[Sponsors] |
June 10, 2020, 11:56 |
rhoPimpleFoam hangs for collated I/O
|
#1 |
New Member
Miguel Castellano
Join Date: Jun 2020
Posts: 4
Rep Power: 6 |
Hi!
So just what it's written on the title, when trying to run rhoPimpleFoam with the collated filaHandler it just hangs. It remains blocked forever. After many many trials and lots of time spent tying to spot the problem I've come to realize it is nothing related to the OpenFOAM version, nor the MPI backend or threading capabilities of the latter. It must merely be a question of how the case is built, since I've run it on other tutorial cases under the same circumstances and it worked smoothly. The workflow is the ordinary: >> blockMesh >> decomposePar -fileHandler collated >> mpirun -np 4 rhoPimpleFoam -parallel -fileHandler collated and BINGO! Build : _f3950763fe-20191219 OPENFOAM=1912 Arch : "LSB;label=32;scalar=64" Exec : rhoPimpleFoam -parallel -fileHandler collated Date : Jun 10 2020 Time : 16:31:01 Host : michu-laptop PID : 279801 I/O : collated (maxThreadFileBufferSize 0) Threading not activated since maxThreadFileBufferSize = 0. Writing may run slowly for large file sizes. --- HANGS IN HERE FOREVER ---- * Note that I set maxThreadFileBufferSize to 0 in $FOAM_ETC/controlDict so you don't tell it is because my MPI implementation doesn't support threading. I've tested both Gcc and Icc compilations with impi and openmpi. Here's the simplified case on github so you can clone it, run it and see it for yourselves. git clone git@github.com:michu5696/ashee.git Since I'm no OpenFOAM expert, could you explain to me what is wrong with this case? Thanks a lot in advance!! |
|
July 7, 2020, 09:59 |
Solved!
|
#2 |
New Member
Miguel Castellano
Join Date: Jun 2020
Posts: 4
Rep Power: 6 |
Ok. For anybody experiencing the same problem. I just figured out. It took me some time and honestly it's the stupidest thing ever. This must be a bug in the implementation of collated output.
For some reason, if you set your dictionary variables from a personal dictionary by adding "#include mydict" the collated fileHandler hangs forever. End of story. |
|
July 20, 2020, 11:13 |
|
#3 |
Senior Member
Joachim Herb
Join Date: Sep 2010
Posts: 650
Rep Power: 22 |
Have you filed a bug at https://bugs.openfoam.org/rules.php ?
|
|
July 21, 2020, 09:18 |
|
#4 |
New Member
Miguel Castellano
Join Date: Jun 2020
Posts: 4
Rep Power: 6 |
||
November 16, 2020, 21:54 |
|
#5 | |
Senior Member
Join Date: Jul 2019
Posts: 148
Rep Power: 7 |
Quote:
Your post was very helpful. I got stuck on the same issue until I saw it. I had to remove the #include from the decompseParDict and enter the number of subdomains manually. The suggested workaround from the bug report did not help me as I got errors too. I am wondering if you managed to resolve your issue while keeping the #include in the decompseParDict file. Thanks. Last edited by Bodo1993; November 17, 2020 at 16:12. |
||
December 28, 2022, 05:06 |
|
#6 |
New Member
Join Date: Nov 2012
Posts: 18
Rep Power: 14 |
I've only deleted foamdatatofluentdict file and restarted the job. It worked...
|
|
Tags |
collated, i/o, rhopimplefoam |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
pimpleFoam runs slower than rhoPimpleFoam | Kosuke Seto | OpenFOAM Running, Solving & CFD | 3 | May 27, 2023 15:12 |
rhoPimpleFoam Boundary Condition Problem | dancfd | OpenFOAM Pre-Processing | 18 | September 16, 2021 08:43 |
rhoPimpleFoam gives stranger result ... and doesn't work | kin3062 | OpenFOAM Running, Solving & CFD | 16 | April 12, 2019 08:46 |
Pressure stair-step behaviour using rhopimplefoam | joegi.geo | OpenFOAM Running, Solving & CFD | 3 | December 12, 2014 13:10 |
rhoPimpleFoam floating point error | dancfd | OpenFOAM Running, Solving & CFD | 6 | January 5, 2014 21:57 |