warning from CPP Driver while inserting row - Unable to find key column 'key' in prepared query

[Prerana Khare]

I have been using Cassandra DB for a while starting from version 2.1.8 to 2.2.1.
I have started getting a warning from CPP driver while using Cassandra 2.2.1

WARN:src/prepared.cpp:54:cass\3a\3aPrepared\3a\3aPrepared(const cass\3a\3aResultResponse*, const std\3a\3astring&, const std\3a\3avector<std\3a\3abasic_string<char, std\3a\3achar_traits<char>, std\3a\3aallocator<char> >, std\3a\3aallocator<std\3a\3abasic_string<char, std\3a\3achar_traits<char>, std\3a\3aallocator<char> > > >&):Unable to find key column 'key' in prepared query

I was not getting this error when using Cassandra version 2.1.8.
Is this any issue with the way we use prepare query?

I need to understand more about this warning so that if required we can eliminate this issue.

More Details:
---------------

Table Definition:
CREATE TABLE if not exists keyspace1.tbl1 (
id bigint,
attrid text,
attrvalue blob,
PRIMARY KEY(id, attrid)
) WITH compaction = { 'class' : 'LeveledCompactionStrategy' };

The warning comes when we use following query string and send it to prepare -

INSERT INTO tbl1 ("id", "attrid", "attrvalue") VALUES (?, ?, ?) USING TIMESTAMP ?

We are getting this error after calling cass_future_get_prepared function

Appreciate help here.

Thanks,
Prerana

[Michael Penick]

Thanks for reporting. I'll need to investigate this a bit more.

Mike

To unsubscribe from this group and stop receiving emails from it, send an email to cpp-driver-us...@lists.datastax.com.

[Prerana Khare]

Thanks Mike. Let me know if you need more information.

[Etienne]

On Thursday, September 24, 2015 at 11:54:14 PM UTC+2, Michael Penick wrote:
> Thanks for reporting. I'll need to investigate this a bit more.
>
>
> Mike
>

I have the same issue; it seems the error comes from the method that tries to get the indices of the key columns. This method relies on a "key_aliases" field in the schema metadata, that does not exist (was it removed with C* 2.2?)

Looking at the sources of the java driver, it should instead rely on the "partition_key" type of the columns metadata. Something like:
for (auto it: columns_) {
if (it.second.get_field("type")->value()->to_string() == "partition_key")
output->push_back(it.first);
}

Etienne

[Michael Penick]

You're right. "key_alisaes" is gone in C* 2.2 and that causes it to incorrectly use the C* 1.2 thrift fallback: https://github.com/datastax/cpp-driver/blob/master/src/schema_metadata.cpp#L533-L545. That's the reason for 'key' in the warning: "Unable to find key column 'key' in prepared query". Going forward the driver should use the primary key metadata provided in the prepared response (https://datastax-oss.atlassian.net/browse/CPP-306) which will be available in the final version of 2.2 release of the driver which should be out in the next few weeks.

Mike

[Prerana Khare]

Hi Mike,

Thanks a lot for the response.
We will wait for the next version.
However it would be great if you can let me know if there is any impact of this issue?

[Michael Penick]

This disables token-aware routing for requests, but wouldn't have any other impact. Token-aware routing allows the driver to routes requests directly to one of the replicas instead of relying on a server-side coordinator node to route the request to a replica. This could potentially increase the latency of requests.

Mike

[Prerana Khare]

Hi Mike,

Ok. Then we will be waiting for this issue fix in next version.

:)