keyspace and column family creationŠhow to use ConsistencyLevel.ALL with creation?

["Hiller, Dean" <De...@nrel.gov>]

The playOrm test suite drops the keyspace and recreates it for tests to wipe out the in-memory or cassandra db.  Today, we successfully ran our test suite on a 6 node cluster.  The one issue I had though was I needed to sleep after keyspace creation and column family creation.  BEFORE that I researched if could increase the write consistency from QUORUM to ALL instead of sleeping WHERE in this special case ALL meant the whole cluster ;) not just the replicated nodes as when you write data.  Looking at the thrift api, this looks like it does not exist though or am I mistaken?

If anyone knows of a method I can call that will not return until all nodes know about the new keyspace and another for the new column family, please do let me know.  (I am hoping that doesn't block data write/read operations of the cluster in general of course).

In our system, we are constantly creating new column families….in fact, it is part of a data upload process and so for now I need to sleep BUT I don't know how long I have to sleep for :( :( :( so a blocking method call would be much better.

Thanks,
Dean

Re: keyspace and column family creationŠhow to use ConsistencyLevel.ALL with creation?

[aaron morton <aa...@thelastpickle.com>]

 describe_schema_versions on the thrift API (SystemManager.describe_schema_versions() on pycassa) is what you are after.

Call it before hand to know what's what, and then call until all nodes converge on the new schema. 

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

On 30/08/2012, at 1:30 AM, "Hiller, Dean" <De...@nrel.gov> wrote:

> The playOrm test suite drops the keyspace and recreates it for tests to wipe out the in-memory or cassandra db.  Today, we successfully ran our test suite on a 6 node cluster.  The one issue I had though was I needed to sleep after keyspace creation and column family creation.  BEFORE that I researched if could increase the write consistency from QUORUM to ALL instead of sleeping WHERE in this special case ALL meant the whole cluster ;) not just the replicated nodes as when you write data.  Looking at the thrift api, this looks like it does not exist though or am I mistaken?
> 
> If anyone knows of a method I can call that will not return until all nodes know about the new keyspace and another for the new column family, please do let me know.  (I am hoping that doesn't block data write/read operations of the cluster in general of course).
> 
> In our system, we are constantly creating new column families….in fact, it is part of a data upload process and so for now I need to sleep BUT I don't know how long I have to sleep for :( :( :( so a blocking method call would be much better.
> 
> Thanks,
> Dean