You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First up: Apologies, I don't have a good example for reproducing the issue, as I use prefect and not dask directly.
TL;DR: New release of kubernetes-asyncio (22.6) broke something in dask-kubernetes (or in prefect, I don't know), pinning version fixed my workflows. Hope this helps someone.
Yesterday, my dask pipelines all worked perfectly in AWS EKS. Today, after re-building images, I consistently got the following:
...
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/dask_kubernetes/core.py", line474, in__init__super().__init__(**self.kwargs)
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/deploy/spec.py", line260, in__init__self.sync(self._start)
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/utils.py", line311, insyncreturnsync(
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/utils.py", line378, insyncraiseexc.with_traceback(tb)
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/utils.py", line351, infresult=yieldfutureFile"/pipelines/flow_test/.venv/lib/python3.9/site-packages/tornado/gen.py", line762, inrunvalue=future.result()
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/dask_kubernetes/core.py", line603, in_startawaitsuper()._start()
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/deploy/spec.py", line293, in_startself.scheduler=awaitself.schedulerFile"/pipelines/flow_test/.venv/lib/python3.9/site-packages/distributed/deploy/spec.py", line59, in_awaitself.start()
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/dask_kubernetes/core.py", line182, instartawaitsuper().start(**kwargs)
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/dask_kubernetes/core.py", line87, instartraiseeFile"/pipelines/flow_test/.venv/lib/python3.9/site-packages/dask_kubernetes/core.py", line77, instartself._pod=awaitself.core_api.create_namespaced_pod(
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/kubernetes_asyncio/client/api_client.py", line192, in__call_apiraiseeFile"/pipelines/flow_test/.venv/lib/python3.9/site-packages/kubernetes_asyncio/client/api_client.py", line185, in__call_apiresponse_data=awaitself.request(
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/kubernetes_asyncio/client/rest.py", line223, inPOSTreturn (awaitself.request("POST", url,
File"/pipelines/flow_test/.venv/lib/python3.9/site-packages/kubernetes_asyncio/client/rest.py", line180, inrequestraiseApiException(http_resp=r)
kubernetes_asyncio.client.exceptions.ApiException: (403)
Reason: ForbiddenHTTPresponseheaders: <...>HTTPresponsebody: {"kind":"Status","apiVersion":"v1","metadata":{},"status":"Failure","message":"pods is forbidden: User \"system:anonymous\" cannot create resource \"pods\" in API group \"\" in the namespace \"prefect\"","reason":"Forbidden","details":{"kind":"pods"},"code":403}
After some investigation, I found that a release of kubernetes-asyncio (version bump 21.7 -> 22.6) has broken dask-kubernetes in all my pipelines.
After adding constraint on version kubernetes-asyncio<22.6, all my pipelines work. Posting this in-case anyone else has similar issues.
The text was updated successfully, but these errors were encountered:
First up: Apologies, I don't have a good example for reproducing the issue, as I use prefect and not dask directly.
TL;DR: New release of kubernetes-asyncio (22.6) broke something in dask-kubernetes (or in prefect, I don't know), pinning version fixed my workflows. Hope this helps someone.
Yesterday, my dask pipelines all worked perfectly in AWS EKS. Today, after re-building images, I consistently got the following:
After some investigation, I found that a release of
kubernetes-asyncio
(version bump 21.7 -> 22.6) has broken dask-kubernetes in all my pipelines.After adding constraint on version
kubernetes-asyncio<22.6
, all my pipelines work. Posting this in-case anyone else has similar issues.The text was updated successfully, but these errors were encountered: