You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@beam.apache.org by "Heejong Lee (Jira)" <ji...@apache.org> on 2019/09/20 23:21:00 UTC

[jira] [Updated] (BEAM-8289) Failing the job when the temporary file is accidentally deleted by another job in FileBasedSink

     [ https://issues.apache.org/jira/browse/BEAM-8289?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Heejong Lee updated BEAM-8289:
------------------------------
    Description: There was an issue that a temporary directory name is not unique in FileBasedSink (https://issues.apache.org/jira/browse/BEAM-7689) and we've changed the naming algorithm from timestamp to UUID. However, it looks like UUID is still not enough to distinguish two separate temporary directories from different jobs. We may need to fail the job when the temporary file is accidentally deleted by another job.  (was: There was an issue that a temporary directory name is not unique in FileBasedSink (https://issues.apache.org/jira/browse/BEAM-7689) and we've changed the naming algorithm from timestamp to UUID. However, it looks like UUID is still not enough to distinguish two separate temporary directories from different jobs. We may need to put job name or job id to the temporary directory name for avoiding future naming collision.)

> Failing the job when the temporary file is accidentally deleted by another job in FileBasedSink
> -----------------------------------------------------------------------------------------------
>
>                 Key: BEAM-8289
>                 URL: https://issues.apache.org/jira/browse/BEAM-8289
>             Project: Beam
>          Issue Type: Bug
>          Components: io-java-files
>            Reporter: Heejong Lee
>            Assignee: Heejong Lee
>            Priority: Major
>
> There was an issue that a temporary directory name is not unique in FileBasedSink (https://issues.apache.org/jira/browse/BEAM-7689) and we've changed the naming algorithm from timestamp to UUID. However, it looks like UUID is still not enough to distinguish two separate temporary directories from different jobs. We may need to fail the job when the temporary file is accidentally deleted by another job.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)