Skip to content

When writing incomplete chunks data gets lost. #39

Description

@odinsbane

I have a zarr folder with too many files. 136000 64x64x64x1x1 chunks. In order to fix this issue I changed the chunk size to 64x64x64x1x1000. Which makes the resulting folder much easier to deal with.

The problem is that I am creating the zarr folder from images stored in TIFF files with N frames each, where N < 1000. I create the zarr file, write the first set of images. Then re-open the zarr file and write the next set of images.

After I am done some of the images are all 0's.

Maybe I am doing something wrong, by opening and closing the writer, I can make a test case, but probably not before August.

In the meantime I am using the zarr file with an excess of files that I generated from the tiff files, so it shouldn't have incomplete chunks to update.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions