Florian Jetter

I can confirm this behavior. I'm not entirely sure who's at fault. I'm not very familiar with this dtype format but to me this looks like a numpy bug (but...

parquet performance will also depend on the backend used. the pyarrow backend is / should be faster but it has still a lot of sharp edges and isn't the default

Well, they are running on a `r6i.16xlarge` machine which has 64 cores. This can't work 😂 The GIL will kill everything. We have moderate if not severe GIL issues already...

They are [submitting a task](https://github.com/bodo-ai/Bodo/blob/879cf934785e2d6370d75cae6d12b702a91c302f/benchmarks/nyc_taxi/dask/nyc_taxi_precipitation.py#L124-L129) that is internally calling [compute](https://github.com/bodo-ai/Bodo/blob/879cf934785e2d6370d75cae6d12b702a91c302f/benchmarks/nyc_taxi/dask/nyc_taxi_precipitation.py#L93). This will be using a worker client, i.e. the driver code is also a worker task. I'm not entirely sure...

I'm not familiar with Ray and can't tell what's necessary to make this work. If the failure is due to the removal of an internal API this is hardly related...

I briefly peaked into the issue. 1. The timeout error is a red herring. The nanny isn't coming up and the `gen_cluster` fixture is retrying for a minute. After the...

> There may be implications for some of the dashboard components, the "pew pew pew" plot comes to mind. I see this is still a draft, let me know when...

I think this is slowly coming together. I see some kind of unexpected failures still in `distributed/shuffle/tests/test_merge.py::test_merge`. They indicate some shuffle failed errors that I cannot see on any test...

Test failures are all over the place. I don't think anything is related.

I triggered another A/B just to make sure