Error message in LIMS environment

Two CellProfiler pipelines are necessary to capture our final readout. The output of the first CellProfiler pipeline is the input for the second one. After loading the input data, the second CellProfiler pipeline results in creating a BatchFile (Batch_data.h5), which is then ran in a batch mode in the cluster-computing context. The pipelines designed in CellProfiler-3.1.8 worked perfectly in GUI or in LIMS environments (with smaller size inputs).

However I get twice an error message with a particular set of images (about 22 000 images). Raw data and the output files from the first pipeline are all in place and not corrupted. Unfortunately running the second pipeline in the same LIMS environment resulted (after 85% completion and twice) in stopping the requested job as depicted in the following screenshot and highlight here:

“autodetected range of [{}, {}] is not finite”.format(first_edge, last_edge))

Few seconds later, the process seems to restart before finally being totally stopped.

Dissecting the input files (for the second pipeline) is not an option as the final readout must be generated as a single .csv file.

Did anyone ever face this error message? Could anyone help me?
Thank you in advance

Is it possible to identify which image “Image #4828” is?

From the error message it looks like it contains an NaN value (or something which has been converted to NaN when read into CellProfiler) which the otsu threshold method doesn’t like.

thank you for your answer
Would say that it is possible,
unfortunately did not yet figured out in which order CellProfiler is processing the files (first by date, then by experiments, then plate, then raw, then column and finally fields???).

all my input files weighted 7Mb, which make me think that it does not come from it.
still remains your hypothesis of a CellProfiler conversion to NaN but note that, among all files processed (in two years) this is the first time it is happening and previous files in the list were processed without any pb until reaching that one which cause CP to stop.
Is there a way to ask CP to skip the problematic files and to go to the next one ?

thank you for your help