If you checkout the latest 4.x or master branch (depending on which
HBase version you use), you can test using transactions in Phoenix.
Tephra is used for transaction support and is used to perform conflict
detection.
To start the Tephra transaction manager you have to set set the
following config in hbase-site.xml on your server
<property>
<name>data.tx.snapshot.dir</name>
<value>DIR_NAME</value>
</property>
and $HBASE_HOME needs to be set before running ./bin/tephra start
To create a transactional table you can use the TRANSACTIONAL attribute for eg:
CREATE TABLE MY_TABLE (k INTEGER PRIMARY KEY, v VARCHAR) TRANSACTIONAL=true;
If you upsert a batch of rows and call commit all or none of the rows
will be committed. A transaction is started implicitly when you
execute a statement over a transactional table. If two clients modify
the same row at the same time an exception will be thrown. Within a
transaction you are also able to see (query) uncommitted data.
You can see more examples of how to use transactions in TransactionIT.java
Transaction support will be in the next version of Phoenix (4.7.0)
which will be released soon.
Thanks,
Thomas
On Thu, Dec 3, 2015 at 11:00 AM, Thomas D'Silva <tdsi...@salesforce.com> wrote:
> Phoenix is working on supporting transactions using tephra (see
> PHOENIX-1674). If you want to use HBase directly with tephra, the
> tephra website has a getting started guide with an example (see
> https://github.com/caskdata/tephra).
>
> -Thomas
>
> On Thu, Dec 3, 2015 at 1:02 AM, Hardika Catur Sapta
> <hardikacatursa...@gmail.com> wrote:
>> Hay guys,
>> We plan to integrate with Apache HBase apache tephra, which we know it uses
>> apache phoenix.
>>
>> How do the operation Tephra? whether using apache phoenix ?? and how to run
>> it.
>>
>> Thank you,
>> -Hardika C. S.
