Closed valeriupredoi closed 1 month ago
indeed reverting to dask == 2024.7.1
and running with --cache-clear
tests pass; dask == 2024.8.0
is the bugger, incoming PR to pin it, then off to create a minimal repro case to tell the Dask folks
the real reason why we see this is this https://github.com/SciTools/iris/issues/6109 and our actual issue that should not be closed until Iris provide a fix https://github.com/ESMValGroup/ESMValCore/issues/2506
Test failures since last night; it appears to be Dask the trouble maker, am investigating right now.