Joe McDonnell has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/13545 )

Change subject: IMPALA-8630: Include partition id when calculating consistent 
remote placement
......................................................................


Patch Set 1:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/13545/1//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/13545/1//COMMIT_MSG@17
PS1, Line 17: This adds the partition_id in the hash, so files with the same
> How are the partition IDs generated? I'm wondering if there might be a less
They are just sequential numbers that we make up, but it turns out that they 
are unique, even across tables:

https://github.com/apache/impala/blob/master/fe/src/main/java/org/apache/impala/catalog/HdfsPartition.java#L572

But that creates a different problem, which is that they are not stable over 
time or across different coordinators. If I invalidate metadata and rerun a 
query, I get different partition ids. I'll explore ways to make something 
stable to use.



--
To view, visit http://gerrit.cloudera.org:8080/13545
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I46c739fc31af539af2b3509e2a161f4e29f44d7b
Gerrit-Change-Number: 13545
Gerrit-PatchSet: 1
Gerrit-Owner: Joe McDonnell <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Joe McDonnell <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-Comment-Date: Mon, 10 Jun 2019 18:31:16 +0000
Gerrit-HasComments: Yes

Reply via email to