I see, thanks a lot!
On Mon, Apr 16, 2012 at 7:41 PM, Ian Varley <[EMAIL PROTECTED]> wrote:
> More complex answer: generally, nothing that involves more than a single
> row in HBase is transactional. :)
> It's possible that HBase might get some limited form of multi-row
> transactions in the future (see HBase-5229<
> https://issues.apache.org/jira/browse/HBASE-5229> for more on that) but
> even then, things would only be transactional within a single region
> server, which means it's not really a general solution for the case you
> mention below (short of some external guarantee that all of your deletes
> are on the same RS).
> That said: mutations are generally idempotent in HBase (except for
> increments). So if you get an exception, it's usually OK to just retry the
> whole thing.
> On Apr 16, 2012, at 6:33 PM, Jean-Daniel Cryans wrote:
> Simple answer: it's not transactional.
> On Mon, Apr 16, 2012 at 4:28 PM, Haijia Zhou <[EMAIL PROTECTED]<mailto:
> [EMAIL PROTECTED]>> wrote:
> Very simple question as the subject shows:
> Is htable.delete(List<Delete>) transactional?
> Say if I am to delete 1000 rows and in the middle of deletion some error
> occurs, then will the whole deletion operation get rolled back or will it
> end up with partial deletion?