Can't open volumetric data via S3 (works locally!)
constantinpape opened this issue · 3 comments
Opening the data generated with https://github.com/ome/ome-ngff-prototypes/tree/main/workflows/mobie-example via s3 does not work. Opening it locally works without issues.
Here's napari when opening it locally via $ napari Covid19-S4-Area2/images/ome-zarr/raw.ome.zarr:
And here's napari when opening it from s3, eitther with
$ napari https://s3.embl.de/i2k-2020/project-ome-zarr/Covid19-S4-Area2/images/ome-zarr/raw.ome.zarr or napari https://uk1s3.embassy.ebi.ac.uk/bia-mobie-test/Covid19-S4-Area2/images/ome-zarr/raw.ome.zarr. (The dataset is available both via the EMBL-Heidelberg and EBI S3):
Note that the data can be accessed from these s3 stores via BigDataViewer (Plugin->BigDataViewer->OME ZARR->Open OME ZARR from S3..., needs MoBIE update site and latest version of it).
@constantinpape running napari -vv I see that the engine is trying to load chunks e.g. https://uk1s3.embassy.ebi.ac.uk/bia-mobie-test/Covid19-S4-Area2/images/ome-zarr/raw.ome.zarr/s0/0/0/0 However there does not seem to be any such object in the bucket which likely explains the black planes.
Runningaws --endpoint-url https://uk1s3.embassy.ebi.ac.uk s3 ls s3://bia-mobie-test/Covid19-S4-Area2/images/ome-zarr/raw.ome.zarr/s0/, it looks like the chunks are . separated rather than / separated. Is that also the case in your local version?
@sbesson you're right, the chunks are . separated. Thanks for noticing. I know why this happens and it's only sort-of my fault...
I will open another issue in ome-zarr-py about this.
(Also interesting that it works locally ....)
Ok, nvmd I confused something, it's completely my fault :D. I will fix it and reupload the data.