[
https://issues.apache.org/jira/browse/HADOOP-19801?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18072591#comment-18072591
]
ASF GitHub Bot commented on HADOOP-19801:
-----------------------------------------
EnricoMi commented on PR #8306:
URL: https://github.com/apache/hadoop/pull/8306#issuecomment-4223921515
@deepujain how did you test this?
Did you run `ITestS3ADeleteNonEmptyDirectoryCapability` against your VAST S3
endpoint?
Did you
1. set `fs.s3a.delete.non-empty-directory.enabled` to `true`?
2. set `fs.s3a.client.s3.custom.headers` to `x-amz-delete-contents=true`?
With 1., obviously, assertions in tests `testDeleteDeepEmptyDir`,
`testDeleteNonEmptyDirRecursive` and `testCapabilityDisabledByDefault` fail.
This is expected because dirs still exist after deletion fails silently.
With 1. and 2., non-delete requests fail for me with `Invalid Argument`
(`fs.delete` involves non-delete requests):
```
org.apache.hadoop.fs.s3a.AWSBadRequestException: List vask-bucket:/test/
delimiter=/ keys=2 requester pays=null on /:
software.amazon.awssdk.services.s3.model.S3Exception: Invalid Argument
(Service: S3, Status Code: 400, Request ID: 31f13100005f19, Extended Request
ID: 31f13100005f19) (SDK Attempt Count: 1):InvalidArgument: Invalid Argument
(Service: S3, Status Code: 400, Request ID: 31f13100005f19, Extended Request
ID: 31f13100005f19) (SDK Attempt Count: 1)
at
org.apache.hadoop.fs.s3a.S3AUtils.translateException(S3AUtils.java:271)
at org.apache.hadoop.fs.s3a.Invoker.retryUntranslated(Invoker.java:481)
at org.apache.hadoop.fs.s3a.Invoker.retryUntranslated(Invoker.java:431)
at
org.apache.hadoop.fs.s3a.S3AFileSystem.listObjects(S3AFileSystem.java:3106)
at
org.apache.hadoop.fs.s3a.S3AFileSystem.s3GetFileStatus(S3AFileSystem.java:4096)
at
org.apache.hadoop.fs.s3a.S3AFileSystem.innerGetFileStatus(S3AFileSystem.java:3987)
at
org.apache.hadoop.fs.s3a.S3AFileSystem$MkdirOperationCallbacksImpl.probePathStatus(S3AFileSystem.java:3846)
at
org.apache.hadoop.fs.s3a.impl.MkdirOperation.probePathStatusOrNull(MkdirOperation.java:216)
at
org.apache.hadoop.fs.s3a.impl.MkdirOperation.getPathStatusExpectingDir(MkdirOperation.java:237)
at
org.apache.hadoop.fs.s3a.impl.MkdirOperation.execute(MkdirOperation.java:134)
at
org.apache.hadoop.fs.s3a.impl.MkdirOperation.execute(MkdirOperation.java:59)
...
```
This is against VAST 5.4.3.1.
> Allow to skip recursive file deletion for non-empty directory
> -------------------------------------------------------------
>
> Key: HADOOP-19801
> URL: https://issues.apache.org/jira/browse/HADOOP-19801
> Project: Hadoop Common
> Issue Type: New Feature
> Components: fs/s3
> Reporter: Enrico Minack
> Priority: Major
> Labels: pull-request-available
>
> A non-empty directory, as well as a directory where its emptiness is unknown,
> is deleted by listing all contained files (or objects with matching key
> prefix), then deleting those files (objects) in bulk delete requests, and
> finally deleting the (then) empty directory itself.
> The recursive deletion of objects is not needed for S3 endpoints that support
> deleting non-empty directories (or path prefixes).
> For example: The [VAST S3 endpoint supports deleting non-empty
> directories|https://kb.vastdata.com/documentation/docs/using-trash-folder-for-s3-objects-6]
> with a single request. Such a feature can only be fully leveraged if
> {{hadoop-aws}} coupd skip the recursive deletion.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
