[
https://issues.apache.org/jira/browse/FLINK-26698?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17508041#comment-17508041
]
Yun Tang commented on FLINK-26698:
----------------------------------
+1, for path like "dfs://schema/path/to", #getPath() would return "/path/to"
which will drop the schema.
> FileSystemJobResultStore#constructDirtyPath might lost the scheme
> -----------------------------------------------------------------
>
> Key: FLINK-26698
> URL: https://issues.apache.org/jira/browse/FLINK-26698
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.15.0, 1.16.0
> Reporter: Yang Wang
> Assignee: Matthias Pohl
> Priority: Blocker
> Fix For: 1.15.0
>
>
>
> {code:java}
> /**
> * Given a job ID, construct the path for a dirty entry corresponding to
> it in the job result
> * store.
> *
> * @param jobId The job ID to construct a dirty entry path from.
> * @return A path for a dirty entry for the given the Job ID.
> */
> private Path constructDirtyPath(JobID jobId) {
> return new Path(this.basePath.getPath(), jobId.toString() +
> DIRTY_FILE_EXTENSION);
> } {code}
>
> Just like above piece of code, we are using {{{}this.basePath.getPath(){}}},
> not directly use {{this.basePath}} when create a new Path. I am afraid this
> will cause scheme lost and cause issue when some filesystem implementation
> tries to stat the path.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
