[
https://issues.apache.org/jira/browse/PHOENIX-2036?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Josh Mahonin updated PHOENIX-2036:
----------------------------------
Attachment: PHOENIX-2036-spark.patch
Call SchemaUtil.getEscapedArgument before setting the input table name in
PhoenixRDD.
Although the only other invocations of this seem to be in integration tests, I
feel like this is the right strategy. First, the unit tests don't have to
change, so the existing API remains consistent. Secondly, I think it follows
the principle of least astonishment. The tableName will be passed exactly as
specified by the RDD/DataFrame, before getting escaped and passed to the the
PhoenixRuntime.
> PhoenixConfigurationUtil should provide a pre-normalize table name to
> PhoenixRuntime
> ------------------------------------------------------------------------------------
>
> Key: PHOENIX-2036
> URL: https://issues.apache.org/jira/browse/PHOENIX-2036
> Project: Phoenix
> Issue Type: Bug
> Reporter: Siddhi Mehta
> Assignee: maghamravikiran
> Priority: Minor
> Attachments: PHOENIX-2036-spark.patch, PHOENIX-2036-v1.patch,
> PHOENIX-2036-v1.patch, PHOENIX-2036-v2.patch, PHOENIX-2036.patch
>
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> I was trying a basic store using PhoenixHBaseStorage and ran into some issues
> with it complaining about TableNotFoundException.
> The table(CUSTOM_ENTITY."z02") in question exists.
> Looking at the stacktrace I think its likely related to the change in
> PHOENIX-1682 where phoenix runtime expects a pre-normalized table name.
> We need to update
> PhoenixConfigurationUtil.getSelectColumnMetadataList(Configuration) be pass a
> pre-normalized table.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
