Hi Nate
On Feb 2, 11:40 pm, Nate McCall <
n...@datastax.com> wrote:
> Hi Lewis,
> I've never actually tried running cassandra with only programatic
> configuration (w/o a yaml file) - is this what you mean?#
Yeah, I think it is. The module was originally written for a rather
dated version
Cassandra and didn't aim to support a yaml file. The aim was to
provide
all configuration in a type of *-site.xml similar to Hadoop and HBase.
The
DataStoreFactory would then pull this out when createStore() was
called.
The proposed resolution is to move away from this, instead opting for
all
the Cassandra config to be pulled from .yaml instead. It would appear
that
this is a more popular way people are working... for obvious reasons.
> I'm wondering if it is even possible given the amount of static
> initializer stuff in DatabaseDescriptor particularly. My curiosity is
> a bit tickled by this, but I'm not going to get a chance to look at
> this until the weekend, unfortunately.
Do you mean whether it is possible to utilise Hector's
DatabaseDescriptor
to read from the cassandra-site.xml? On first hearing, it sounds like
a nasty
case of using something which is good at it's job (reading config from
yaml) to
do something different (read from *-site.xml) Although I'm curious
with this
as well I'm sceptical about continuing with the cassandra-site.xml, it
just
seems pretty dated to me.
> Several other folks here have some deep experience with init-ing
> cassandra in-process for testing (and other stuff) as well though.
Thanks. Well hopefully my comments above have caused some ripples
which merit feedback.
Thanks again
Lewis