Sadu,
I am using JSON as the input format, with the JSON SerDe from
https://github.com/rcongiu/Hive-JSON-Serde.
A sample JSON record is: (in actual use each JSON record must be on one line
only).
{
"field1":"hello",
"field2":123456,
"field3":1234.5678,
"field4":true,
"field5":{"field5a":"embedded 1", "field5b":44, "field5c":4.44,
"field5d":false, "field5e":[12,13,14]},
"field6":[2, 3, 4, 5, 6],
"field7":[4.33, 5.33, 6.33],
"field8":["one", "two", "three"],
"field9":[[1,2,3], [4,5], [6,7,8]],
"field10":[["smith","jones"], ["bob", "bill"]],
"field11":[{"f11a":"one", "f11b":"two"}, {"f11a":"three", "f11b":"four"}]
}
My table definition is: (ignore the fact that the fields are listed out of
order, this does not matter)
CREATE TABLE tt1
(
field8 ARRAY<STRING>,
field9 ARRAY<ARRAY<INT>>,
field2 INT,
field3 DOUBLE,
field1 STRING,
field6 ARRAY<INT>,
field7 ARRAY<DOUBLE>,
field4 BOOLEAN,
field5 STRUCT
<
field5d:BOOLEAN,
field5e:ARRAY<INT>,
field5a:STRING,
field5b:INT,
field5c:DOUBLE
>,
field10 ARRAY<ARRAY<STRING>>,
field11 ARRAY<STRUCT
<
f11a:STRING,
f11b:STRING
>>
)
ROW FORMAT SERDE 'org.openx.data.jsonserde.JsonSerDe'
WITH SERDEPROPERTIES ("ignore.malformed.json" = "true")
STORED AS TEXTFILE;
This small sample actually works great! The problem is when I try to scale up
to larger more complex tables.
(Note the latest version of this SerDe has a bug related to number formats. You
should use 1.1.3 and use only INTs and DOUBLEs.)
Chuck
From: Sadananda Hegde [mailto:[email protected]]
Sent: Thursday, October 11, 2012 11:47 PM
To: [email protected]
Subject: Re: Any advice about complex Hive tables?
Hi Chuck,
I have a similar complex hive tables with many fields and some are nested like
array of structs (but only upto 3 levels). How did you define you ROW FORMAT as
to separate the items? The COLLECTION ITEMS TERMINATED BY clause works only for
the first level.How did you handle level 2 , 3, etc? Is it through SERDE
FORMATs? Could you
share your CREATE TABLE statement? I am having problem correctly defining my
DDL to load the data file correctly.
Much appreciated.
Thanks,
Sadu
On Mon, Oct 8, 2012 at 3:37 PM, Connell, Chuck
<[email protected]<mailto:[email protected]>> wrote:
(Follow up to the thread below...)
I have a complex Hive table -- many fields, many nested structs. Hive fails to
create the table at all. I can't even start to load data or query the data.
Anyone else run into this? It seems to be a showstopper to using Hive for us,
which is a shame since everything else about Hive is just what we want.
Any suggestions? Workarounds?
Thanks,
Chuck
________________________________________
From: Connell, Chuck
Sent: Thursday, October 04, 2012 4:31 PM
To: [email protected]<mailto:[email protected]>
Subject: RE: Limit to columns or nesting of Hive table?
The issue apparently is not just the number of levels of nesting. I just
created a Hive table with 20 levels of structs within each other. It created
fine. This is more levels than the table that was failing for me. The failing
table had many more fields throughout the levels.
Chuck
-----Original Message-----
From: Connell, Chuck
[mailto:[email protected]<mailto:[email protected]>]
Sent: Thursday, October 04, 2012 12:09 PM
To: [email protected]<mailto:[email protected]>
Subject: RE: Limit to columns or nesting of Hive table?
Thanks. So is the nesting limit 10 now? Does your 2nd paragraph mean that this
limit cannot easily be raised?
Chuck
-----Original Message-----
From: Edward Capriolo
[mailto:[email protected]<mailto:[email protected]>]
Sent: Thursday, October 04, 2012 11:57 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: Limit to columns or nesting of Hive table?
There is an open jira ticket on this. There is a hard coded limit but it could
be raised with some mostly minor code changes.
One of the bigger problems is that hive stores the definition of a column in
JDBC "column" and for some databases larger nested structs can case issues.
Edward
On Thu, Oct 4, 2012 at 11:48 AM, Connell, Chuck
<[email protected]<mailto:[email protected]>> wrote:
> I am trying to create a large Hive table, with many columns and deeply
> nested structs. It is failing with java.lang.ArrayIndexOutOfBoundsException:
> 10.
>
>
>
> Before I spend a lot of time debugging my table declaration, is there
> some limit here I should know about? Max number of columns? Max depth
> of struct nesting?
>
>
>
> Thanks,
>
> Chuck
>
>
