View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002835 | OpenFOAM | Bug | public | 2018-02-08 10:50 | 2018-02-13 16:53 |
| Reporter | Joris.C | Assigned To | henry | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Fixed in Version | dev | ||||
| Summary | 0002835: decomposePar -force fails to remove time folders in processor* dirs | ||||
| Description | When I have a case which has already run in parallel, and thus a time folder (eg 0.156) in the processor-subdirs, running decomposePar on the case (with only a 0 time folder) fails to remove these time folders. The specific commands I am runnning: decomposePar -case xx -force -allRegions decomposePar -case xx -region yy -force Neither of them remove this time folder. Also when I change the number of processors, it fails to remove this time folder and/or processor dirs (when I lower the number of processors). | ||||
| Steps To Reproduce | Run a case in parallel and have it write a time, decompose again afterwards. | ||||
| Additional Information | It seems to me this was introduced by v5. | ||||
| Tags | No tags attached. | ||||
|
|
I ran the cavity case in parallel on 2 processors and then changed the numberOfSubdomains to 4 and ran decomposePar -force and it worked fine and the case ran fine on 4 processors. Could you provide the steps to reproduce the problem on one of the tutorial cases? |
|
|
I ran the case on 4 processors then reset the numberOfSubdomains to 2 and ran decomposePar -force and it worked fine and the case ran fine on 2 processors. |
|
|
1) Run OpenFOAM-5.x/tutorials/heatTransfer/chtMultiRegionFoam/multiRegionHeater with ./Allrun 2) Interrupt 3) Check whether there is a time folder in processor0 (for me it was 10) and none in the case folder itself (for me it has reconstructed, but I removed it manually) 4) decomposePar -allRegions -force, which should decompose time 0 5) Time folder 10 is still in processor folder Giving that it works fine for the cavity case, it is probably connected to the allRegions argument. |
|
|
Currently forced removal of the previous decomposition is disallowed for specified region and all regions: if (region || allRegions) { procDirsProblem = false; forceOverwrite = false; } It is not clear why this should be disallowed is it is a command line option and need not be specified if not needed. I will run some tests. |
|
|
commit 6a417d4a931f406e6e2875135134def052ee6763 Note that this adds support for -force with -allRegions but for -region which would be more complicated as it requires seaching the processor directory trees for region sub-directories. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2018-02-08 10:50 | Joris.C | New Issue | |
| 2018-02-08 11:17 | henry | Note Added: 0009280 | |
| 2018-02-08 11:18 | henry | Note Added: 0009281 | |
| 2018-02-08 12:22 | Joris.C | Note Added: 0009283 | |
| 2018-02-13 10:31 | henry | Note Added: 0009308 | |
| 2018-02-13 16:53 | henry | Assigned To | => henry |
| 2018-02-13 16:53 | henry | Status | new => resolved |
| 2018-02-13 16:53 | henry | Resolution | open => fixed |
| 2018-02-13 16:53 | henry | Fixed in Version | => dev |
| 2018-02-13 16:53 | henry | Note Added: 0009310 |
