iam using the cdh 5.11 and my catalog server is not starting even i keeping the Impala Catalog Server Environment Advanced Configuration Snippet (Safety Valve) as JAVA_TOOL_OPTIONS=-Xmx8g or JAVA_TOOL_OPTIONS=-Xss2m ... View more

Thanks ... View more

Hi Shannon,   Impala does not yet support array or map subscripts. What you can do is something like this:   select a.value from mytable t, t.myarray a where a.pos in (0,1)   But this will give you the "values" elements as rows and not in different columns.     ... View more

Thanks, that worked. ... View more

That's an interesting scenario. Unfotunately, there is no elegant way to do what you ask in Impala's SQL.   The query you really want to write is this one: select * from (select *, parts from example ex order by value desc limit 2) v, v.parts   The issue is that Impala currently does not support returning complex types in the select-list of an inline view: https://issues.apache.org/jira/browse/IMPALA-2777   Note that '*' in Impala expands to all scalar-typed colmns, so you need to list 'parts' explicitly. ... View more

https://issues.apache.org/jira/browse/IMPALA-5243 ... View more

Since CDH 5.8 you can use  set PARQUET_FALLBACK_SCHEMA_RESOLUTION=1; to look up columns within Parquet files by column name. For more, see: https://www.cloudera.com/documentation/enterprise/latest/topics/impala_parquet_fallback_schema_resolution.html     ... View more