[
https://issues.apache.org/jira/browse/PHOENIX-4678?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16418032#comment-16418032
]
Josh Elser commented on PHOENIX-4678:
-------------------------------------
{quote}How it happens that it's not null?
{quote}
So, it is null at this point. We eventually pass this null schema into the
construction of the PTable which creates an empty PName instead. This is then
passed into SchemaUtil.getQualifiedTableName via IndexColumnNames.
> IndexScrutinyTool generates malformed query due to incorrect table name(s)
> --------------------------------------------------------------------------
>
> Key: PHOENIX-4678
> URL: https://issues.apache.org/jira/browse/PHOENIX-4678
> Project: Phoenix
> Issue Type: Bug
> Reporter: Josh Elser
> Assignee: Josh Elser
> Priority: Critical
> Fix For: 4.14.0, 5.0.0
>
> Attachments: PHOENIX-4678.diff
>
>
> {noformat}
> HADOOP_CLASSPATH="/usr/local/lib/hbase/conf:$(hbase mapredcp)" hadoop jar
> /usr/local/lib/phoenix-5.0.0-SNAPSHOT/phoenix-5.0.0-SNAPSHOT-client.jar
> org.apache.phoenix.mapreduce.index.IndexScrutinyTool -dt J -it
> INDEX1{noformat}
> This ends up running queries like {{SELECT ... FROM .J}} and {{SELECT ...
> FROM .INDEX1}}.
> This is because SchemaUtil.getQualifiedTableName is not properly handling an
> empty schema name, only a null schema name.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
