Details
-
Bug
-
Status: Triage Needed
-
P2
-
Resolution: Fixed
-
None
-
None
Description
We seem to be able to stage artifacts and retrieve the manifest fine, but retrieving the artifacts doesn't work. This happens on both my k8s Flink cluster and on my local Flink cluster. At a quick glance the artifact is in the jar where it should be. cc robertwb
2019-11-21 18:43:39,336 INFO org.apache.beam.runners.fnexecution.artifact.AbstractArtifactRetrievalService - GetArtifact name: "pickled_main_session"
retrieval_token: "BEAM-PIPELINE/pipeline/artifact-manifest.json"
failed
java.io.IOException: Unable to load e1d24d848414cecf805a7b5c2b950c6430c20eb32875dac00b40f80f3c73a141/ea0d10d07f4601782ed647e8f6ba4a055be13674ab79fa0c6e2fa44917c5264c with org.apache.flink.runtime.execution.librarycache.FlinkUserCodeClassLoaders$ChildFirstClassLoader@785297ac
Attachments
Issue Links
- is related to
-
BEAM-8512 Add integration tests for Python "flink_runner.py"
- Resolved
- links to