[
https://issues.apache.org/jira/browse/KUDU-1159?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jean-Daniel Cryans resolved KUDU-1159.
--------------------------------------
Resolution: Fixed
Fix Version/s: n/a
Target Version/s: (was: 0.8.0)
Resolving this, Todd told me it was done a while ago. In any case, it's more of
an Impala issue than a Kudu issue.
> Impala sinks should validate column indexes after opening table
> ---------------------------------------------------------------
>
> Key: KUDU-1159
> URL: https://issues.apache.org/jira/browse/KUDU-1159
> Project: Kudu
> Issue Type: Bug
> Components: impala
> Affects Versions: Private Beta
> Reporter: Todd Lipcon
> Assignee: David Alves
> Priority: Critical
> Fix For: n/a
>
>
> Through the API, I dropped a column (the last one) from a Kudu table, and
> then issued an update of the dropped column. Apparently, the sink was still
> using a column index calculated based on the impala-side table descriptor,
> rather than looking up the column index from the schema seen when it opened
> the table. This caused it to call set() with an invalid index and crash
> impalad.
> This can probably be reproed without an ALTER TABLE by just creating a
> mapping table that doesn't actually match the underlying schema, and then
> trying to perform an insert or update.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
