If you use the --debug flag when you start the CLI it always will print full 
stack traces. 

What is the CF definition ?  I'm guessing the column_metadata specifies that 
the age column is a Long

Was there existing data in the age column and if so how was it encoded ? Was 
the existing data was encoded as a variable length integer value? The standard 
IntegerType is not compatible with the LongType as the the long is fixed width. 
If this is the case try re-creating the index using an IntegerType. 

This worked for me…

[default@dev] create column family User
...         with comparator = AsciiType
...         and column_metadata =
...         [{
...             column_name : age,
...             validation_class : LongType,
...             index_type : 0,
...             index_name : IdxAge},
...         ];
2fd1a5c0-352b-11e1-0000-242d50cf1fb6
Waiting for schema agreement...
... schemas agree across the cluster
[default@dev] 
[default@dev] get User where age = 1;  

0 Row Returned.
Elapsed time: 33 msec(s).
[default@dev] 

Hope that helps. 

-----------------
Aaron Morton
Freelance Developer
@aaronmorton
http://www.thelastpickle.com

On 31/12/2011, at 6:09 AM, Sasha Dolgy wrote:

> as per the wiki link you sent, i change my query to:
> 
> get user where something = '1';
> 
> Still throws the error ... This was fine *before* I ran the update CF
> command ..
> 
> To Query Data
> get User where age = '12';
> 
> On Fri, Dec 30, 2011 at 6:05 PM, Moshiur Rahman <moshi.b...@gmail.com> wrote:
>> I think you need to mention data type in your command. You have to run the
>> following command first:
>> assume <CFName> keys as <TypeName, i.e., utf8>
>> 
>> Otherwise, you need to mention type with each command, e.g.,
>> utf8('keyname').
>> http://wiki.apache.org/cassandra/CassandraCli
>> 
>> Moshiur
>> 
>> 
>> 
>> On Fri, Dec 30, 2011 at 10:50 AM, Sasha Dolgy <sdo...@gmail.com> wrote:
>>> 
>>> Hi Everyone,
>>> 
>>> Been a while .. without any problems.  Thanks for grinding out a good
>>> product!  On 1.0.6, I applied an update to a column family to add a
>>> secondary index, and now via the CLI, when I perform a "get user where
>>> something=1" I receive the following result:
>>> 
>>> org.apache.cassandra.db.marshal.MarshalException: A long is exactly 8
>>> bytes: 1
>>> 
>>> This behaviour doesn't seem to be affecting phpcassa or hector
>>> retrieving the results of that query ... is this a silly something
>>> i've done, or something a bit more buggy with the CLI?
>>> 
>>> Thanks in advance,
>>> -sd

Reply via email to