I am using scanpy tools on an anndata matrix. When I use for example filter with scanpy or scanpy normalise data, the tool shortly runs and is then marked green, as if it were finished. The next tool in my workflow will then start to run.
When I however open the green tool, it says queued. When I open the Dataset Details, it says
“This programme requires a UTF-8 locale, please check your $LANG setting.”.
It seems the anndata matrix is not there/faulty, because visualization tools do not work.
If I re-run the tool, it sometimes works and gives me a proper output. This seems however random, and take quite a number of manually clicking re-run. Changing the version of the tool doesnt make a difference.
Do you mean that the top level folder of the collection is green in color, but the datasets inside the folder are in different states? If so, that can be expected. Let the jobs inside the collection complete. Deleting and rerunning just starts that whole process again.
A workflow will run any downstream tools on data output from any successful upstream jobs. So, you could have all three states: successful, still processing, and still queued jobs within different collections at any point while the workflow is still processing.
If you think that a downstream tool is trying to execute on data that isn’t processed yet, that would be a problem. The tool would eventually fail if that is true (turn red) and report a message about the inputs not being “ready” when the job executed. Is that happening? At the EU server? Could you share an example or some screenshots?
The top level folder is green, but the dataset inside is queued.
Unfortunately, also visualization tools downstream are already starting when the previous datset is marked green (but still queued), and thus fail and are marked red.
Similar things happen with other downstream tools, if I start them seperately to my own workflow.
@ln1039 Can you try again? It should be working now.