The suggest range may be larger than end row in this region (For example
,nearby the splited row). So to make sure the result is correct, I need
other region to also do the search.
I am using zookeeper for lock just for learning whether my theory will
work , and how many time it will cost. It's n
bq. to get an suggest range search to client, and do this suggested range
search
Since you're using coprocessor, the coprocessor can act (range search)
using the suggestion and return the result to the client.
There is no need to return intermediate result to client.
BTW using zookeeper for locki
In fact , I am using coprocessor to Implement geography index. Each
region will only have part of the Index to index the data that it contained
. When I try to implement KNN search, only one region will contain the
index according to the KNN center point. So this region will give an
suggest
Unsubscribe
Sent from ProtonMail mobile
Original Message
On Jan 15, 2018, 10:49 AM, Ted Yu wrote:
> bq. need region2 to search fist, because it has my own index.
>
> Looks like only some of the regions have your index. Can you tell us more
> about the characteristics of the reg
bq. need region2 to search fist, because it has my own index.
Looks like only some of the regions have your index. Can you tell us more
about the characteristics of the region(s) where index would be present.
bq. My scan will be blocked for my lock on region1,
By 'my' I assume the lock is placed
Hello Everyone
I want to know If the Scan is send to HBase by sequence? For
example, If I have a table that has two region（region1,region2）, and I sent
a Scan() to this Table. This Scan will be send to region1 first and until
region1 returned, region2 can get this Scan? So If There are som
