CFD Online Discussion Forums - Reconstructing cyclicAMI fails

Hey,

I'm having problems reconstructing my case with periodic bc (cyclicAMI). The mesh was imported with fluent3dmeshtofoam. It's got some areas where the periodic faces don't match very well, so I needed to run the case with cyclicAMI Interface and 'lowWeightCorrection 0.2'. No problems running the case in non-parallel, but decomposing it fails at the reconstruction of the processors. This is probably because of the bad cells, where the weight is too low:

Code:

AMI: Patch source sum(weights) min/max/average = 0, 1.00154, 0.997058

AMI: Patch source identified 8 faces with weights less than 0.2

AMI: Patch target sum(weights) min/max/average = 0, 1.00066, 0.996335

AMI: Patch target identified 15 faces with weights less than 0.2

The case runs fine, even in parallel, just the reconstruction fails :(
preserving the patches doesn't do the trick, unfortunately.

Any suggestions for this kiind of problem?
I'M using OF2.3.1