Details
-
Bug
-
Status: Closed
-
Blocker
-
Resolution: Fixed
-
None
-
None
Description
The issue now is exclusive to running on YARN with s3a:// as your configured FileSystem. If so, the Flink session will fail on staging itself because it tries to copy the flink/lib directory to S3 and the S3aFileSystem does not support recursive copy.
Old Issue
Using the RocksDBStateBackend with semi-async snapshots (current default) leads to an Exception when uploading the snapshot to S3 when using the S3AFileSystem.
AsynchronousException{com.amazonaws.AmazonClientException: Unable to calculate MD5 hash: /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886 (Is a directory)}
at org.apache.flink.streaming.runtime.tasks.StreamTask$AsyncCheckpointThread.run(StreamTask.java:870)
Caused by: com.amazonaws.AmazonClientException: Unable to calculate MD5 hash: /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886 (Is a directory)
at com.amazonaws.services.s3.AmazonS3Client.putObject(AmazonS3Client.java:1298)
at com.amazonaws.services.s3.transfer.internal.UploadCallable.uploadInOneChunk(UploadCallable.java:108)
at com.amazonaws.services.s3.transfer.internal.UploadCallable.call(UploadCallable.java:100)
at com.amazonaws.services.s3.transfer.internal.UploadMonitor.upload(UploadMonitor.java:192)
at com.amazonaws.services.s3.transfer.internal.UploadMonitor.call(UploadMonitor.java:150)
at com.amazonaws.services.s3.transfer.internal.UploadMonitor.call(UploadMonitor.java:50)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
Caused by: java.io.FileNotFoundException: /var/folders/_c/5tc5q5q55qjcjtqwlwvwd1m00000gn/T/flink-io-5640e9f1-3ea4-4a0f-b4d9-3ce9fbd98d8a/7c6e745df2dddc6eb70def1240779e44/StreamFlatMap_3_0/dummy_state/47daaf2a-150c-4208-aa4b-409927e9e5b7/local-chk-2886 (Is a directory)
at java.io.FileInputStream.open0(Native Method)
at java.io.FileInputStream.open(FileInputStream.java:195)
at java.io.FileInputStream.<init>(FileInputStream.java:138)
at com.amazonaws.services.s3.AmazonS3Client.putObject(AmazonS3Client.java:1294)
... 9 more
Running with S3NFileSystem, the error does not occur. The problem might be due to HDFSCopyToLocal assuming that sub-folders are going to be created automatically. We might need to manually create folders and copy only actual files for S3AFileSystem. More investigation is required.
