https://community.cloudera.com/t5/Support-Questions/IMPALA-scale-that-does-not-match-the-table-metadata-scale/m-p/277366#M207540 <P><a href="https://community.cloudera.com/t5/user/viewprofilepage/user-id/24734">@tmater</a>&nbsp;Thank you for the update.</P><P>What if some one&nbsp;alters the table, without knowing this fact that it would not work in impala.</P><P>If we have a huge data set(historical data) that the&nbsp;user&nbsp;will be&nbsp;unable to view&nbsp;in impala.</P><P>what is best way to overcome this.</P><P>1.Is there a way we can edit the parquet files metadata&nbsp;in impala either thru spark/from command line&nbsp;?.</P><P>2.we tried creating a temporary&nbsp;table with new schema&nbsp;and copying data from old table with Lateral view explode(on array/complex data type) in hive(manually with hive queries) and it is&nbsp;taking lot of our cluster&nbsp;resource and terminating without doing anything.</P><P>3.Even we tried to split the historical data(source) with date range and it is not helping us.</P><P>&nbsp;</P><P>I believe Lateral view explode is taking lot of resource on a historical data and causing to terminate.</P><P>&nbsp;</P><P>Notes: we have complex data type in our source table and so we are using Lateral view explode in hive</P><P>&nbsp;</P><P><STRONG>Any work around either in spark/hive/impala anything that will help to deal with historical and current data in impala would be really helpful.</STRONG></P><P>&nbsp;</P><DIV class="dcxa-lithium-author-name">Thank you for your help on this</DIV> Mon, 16 Sep 2019 16:30:44 GMT ravikumashi 2019-09-16T16:30:44Z