View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002988 | OpenFOAM | Feature | public | 2018-06-20 19:44 | 2018-06-29 11:40 |
| Reporter | fertinaz | Assigned To | henry | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Platform | Blue Gene Q | OS | CNK | OS Version | (please specify) |
| Summary | 0002988: timeVaryingMappedFixedValue cannot read boundaryData | ||||
| Description | I am working on the tutorial case "pitzDailyExptInlet" to check "timeVaryingMappedFixedValue" BC and it fails because of the error: "Cannot find starting sampling values for current time 1 -- Have sampling values for times 0()" So I tracked down the issue to: bool foundTime = mapperPtr_().findTime ( sampleTimes_, startSampleTime_, this->db().time().value(), lo, hi ); in function checkTable() in timeVaryingMappedFixedValueFvPatchField.C. This flag returns false. As reported above this is a BG/Q cluster. Login nodes have Power 7 CPUs with CentOS which can run this tutorial case successfully. However, compute nodes have A2 processors with CNK which offloads all the I/O operations to the IO nodes. I suspect that bool foundTime = mapperPtr_().findTime requires modification but cannot seem to find where to make those modifications. Can you point me to the source code? | ||||
| Steps To Reproduce | cd $FOAM_TUTORIALS/incompressible/simpleFoam/pitzDailyExptInlet blockMesh simpleFoam | ||||
| Tags | No tags attached. | ||||
|
|
Are you running with a different fileHandler option than the default? The relevant method for you to investigate (i.e., the function returning false) is Foam::pointToPointPlanarInterpolation::findTime, which can be found at line 361 of the file src/meshTools/triSurface/triSurfaceTools/pointToPointPlanarInterpolation.C. |
|
|
Hello Will, thanks for your respond. 1. I am using the default fileHandler. 2. I already found the correct function however afterwards I realized that the real problem is in POSIX.C: // Read and parse all the entries in the directory while ((list = ::readdir(source)) != nullptr) { // Original OpenFOAM code: fileName fName(list->d_name); On BGQ compute nodes, list->d_name returns 1 character longer than it should be with 0th entry being a non-ascii char. I implemented a workaround for that but you can consider implementing a more flexible solution. |
|
|
This sounds like a bug in the BG/Q OS rather than OpenFOAM, have you reported it to IBM? |
|
|
Apparently a bug in the BG/Q OS and has been or should be reported to IBM. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2018-06-20 19:44 | fertinaz | New Issue | |
| 2018-06-25 16:25 | will | Note Added: 0009822 | |
| 2018-06-27 20:11 | fertinaz | Note Added: 0009837 | |
| 2018-06-27 21:02 | henry | Note Added: 0009838 | |
| 2018-06-27 21:12 | henry | Note Edited: 0009838 | |
| 2018-06-29 11:40 | henry | Assigned To | => henry |
| 2018-06-29 11:40 | henry | Status | new => closed |
| 2018-06-29 11:40 | henry | Resolution | open => no change required |
| 2018-06-29 11:40 | henry | Note Added: 0009841 |
