No, setting the hbase.regionserver.wal.codec config param won't have an impact on other non Phoenix users of your cluster, and yes, you need to restart your cluster after making this config change.

No need for you to do anything regarding the phoenix.query.rowKeyOrderSaltedTable config parameter. If you set the SALT_BUCKETS property at CREATE TABLE time to salt your table, it will be pre-split automatically for you.

BTW, best to continue using our github mailing list until we've switched completely over to Apache (hopefully early next week - we're waiting on a couple of things).


On Wed, Jan 15, 2014 at 12:27 AM, Li Li <> wrote:
sorry, there is another question
will set hbase.regionserver.wal.codec =
affect the performance of other guys who use hbase(not using phoenix)?

On Wed, Jan 15, 2014 at 4:22 PM, Li Li <> wrote:
> btw, to enable mutable index. also need modify hbase-site.xml
> <property>
>   <name>hbase.regionserver.wal.codec</name>
>   <value>org.apache.hadoop.hbase.regionserver.wal.IndexedWALEditCodec</value>
> </property>
> do this need restart region servers?
> On Wed, Jan 15, 2014 at 3:39 PM, Li Li <> wrote:
>> This Rowkey order scan can be turned on by setting the
>> phoenix.query.rowKeyOrderSaltedTable config property to true in your
>> hbase-sites.xml. When set, we disallow user specified split points on
>> salted table to ensure that each bucket will only contains entries
>> with the same salt byte. When this property is turned on, the salted
>> table would behave just like a normal table and would return items in
>> rowkey order for scans.
>> I don't providing split points on creating table, should I still
>> modify hbase-site.xml in all region servers(I should ask admin for
>> help to modify and restart hbase)?