Closed abprime closed 8 months ago
Seems like that might have been from https://github.com/dask/distributed/pull/6700. That's now requiring a new Cluster.wait_for_workers
method, that isn't on GatewayCluster
.
We could implement that (PR would be great). It might be worth opening an issue on dask/distributed to confirm whether that change to the cluster interface was intention (it kind of looks incidental to the intent of the PR, but I haven't looked closely).
This was an upstream issue fixed in dask/distributed#8441 part of distributed>=2024.1.0
, so use of that version should resolve this I think.
Describe the issue: The new version
2023.9.0
is giving an attribute error forwait_for_workers
. This was working in the earlier version2023.1.1
. The method on the cluster is called internally from the distributed client methodwait_for_workers
.Is there any alternative way to wait for the workers?
Minimal Complete Verifiable Example:
Anything else we need to know?:
Environment: