incubator-cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nicolas Romanetti <nromane...@jaxio.com>
Subject Re: digest query: why relying on value?
Date Mon, 02 Apr 2012 14:38:05 GMT
Right on spot thanks!

It would be interesting to have some metrics on how rare is the case:

        // break ties by comparing values.
        if (timestamp() == column.timestamp())
            return value().compareTo(column.value()) < 0 ? column : this;

If extremely rare, it would be may be more efficient to not hash the value
and reclaim it only when hitting a such case (ok easy to say :-))




On Mon, Apr 2, 2012 at 4:25 PM, Jonathan Ellis <jbellis@gmail.com> wrote:

> Look at Column.reconcile.
>
> On Mon, Apr 2, 2012 at 9:17 AM, Nicolas Romanetti <romanetti@gmail.com>
> wrote:
> >  Hello,
> >
> > Why does the digest read response include a hash of the column value?
> Isn't
> > the timestamp sufficient?
> >
> > May be an answer:
> > Is the value hash computed to cope with (I presume rare) race condition
> > scenario where 2 nodes would end up with same col. name and same col.
> > timestamp but with a different col. value ?
> > But then I wonder how to decide which value wins!
> >
> > Sincerely,
> >
> > Nicolas.
>
>
>
> --
> Jonathan Ellis
> Project Chair, Apache Cassandra
> co-founder of DataStax, the source for professional Cassandra support
> http://www.datastax.com
>



-- 
Nicolas Romanetti
06 18 65 03 89
twitter: @nromanetti
http://www.jaxio.com/
http://www.springfuse.com/

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message