How to query to local partition cache data

[dark <ek...@gmail.com>]

How can I query only the data held by ignite data nodes? 
I've used it like this, but I've confirmed that the CompletedTaskCount of
the GridQueryExecutor of other nodes goes up.

ignite.compute().affinityRun(cacheName, 1, () -> {
				Ignite ignite = Ignition.ignite();
				IgniteCache cache = ignite.cache(cacheName);

				SqlFieldsQuery query = new SqlFieldsQuery("select path, sum(count) from
CacheTable group by path").setLocal(true);

				long queryStartTimeMillis = System.currentTimeMillis();

				List all = cache.query(query).getAll();
});

I think that when I query like above, Compute Task is delivered to the node
that is in charge of 1 partition key and queries about the partitions that
it is responsible for. However, the CompletedTaskCount of the
GridQueryExecutor on all nodes has been increased. :(

ignite.compute().affinityRun(cacheName, 1, () -> {
				Ignite ignite = Ignition.ignite();
				IgniteCache cache = ignite.cache(cacheName);

				SqlFieldsQuery query = new SqlFieldsQuery("select path, sum(count) from
CacheTable group by path").setPartitions(1);

				long queryStartTimeMillis = System.currentTimeMillis();

				List all = cache.query(query).getAll();
});

when I query like above, Only the node in charge of the partition has
confirmed that the CompletedTaskCount of the GridQueryExecutor increases.

How can I force the Data node to query with its own data?






--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to query to local partition cache data

[dark <ek...@gmail.com>]

My answer of question describes below code. 

Q. How can I query only the data held by Ignite data nodes? 

A. First, Ignite AffinityFunction is used to send the desired data to one
node with partition 1. And, if you are guaranteed to be assigned to 1
partition, use the code below. Use Ignite AffinityRun | affinityCall to
assign compute task to partition 1 node. The node then assigns a partition
to the setPartitions method to query only the partitions it is responsible
for. This part is described in the code below.

https://github.com/gridgain/gridgain-advanced-examples/blob/master/src/main/java/org/gridgain/examples/datagrid/query/ScanQueryExample.java

@Andrew MashenkovReply Thanks for your kind reply again. 




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to query to local partition cache data

[dark <ek...@gmail.com>]

Below is the best way to do a query based on local data I want. Is this part
right?

When we do affinityCall, we specify partition 1 and the query is also set to
1.

SqlFieldsQuery query = new SqlFieldsQuery ("select sum (count) from data
group by id") setPartitions (1);

Even if we have a backup, is it only based on ignite partition 1 node own
data?




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to query to local partition cache data

[dark <ek...@gmail.com>]

Hi, 

Thanks for kind of reply.

But, I test each query option.

setLocal - In my opinion, it seemed that all the nodes were performing on
the query. What I wanted to do was that the node responsible for the
affinityRun key query only the data. However, as described above, it seemed
to load all nodes.

setCollocated - This part seems to mean that the node requested by
affinityRun becomes the Reducer Node and that the query is specified to have
Group By on each node when generating the Map query. 
In other words, if you do not use this option, it is understood that the
requesting node will ask all nodes for a row, and if they receive all of
them, it will perform a Group By query on their own. Is this part right?

setDistributedJoins - I have not tested this part. We are not using Join. I
understood what you explained. Thank you for your kind explanation.




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to query to local partition cache data

[Andrey Mashenkov <an...@gmail.com>]

Hi,

setLocal - if true then query will be executed on local data node only.
setCollocated - if true then rows are collocated on GROUP BY field and
Ignite can evaluate aggregate functions on Map step.
setDistributedJoins - if true then Ignite will try to join rows with rows
on other nodes.

Would you please clarify what and how you observe the weird things?

BTW:
In first case, affinityRun makes task to be executed on node that is
primary to Key given, but local SQL query will run on all primary data on
available on local node.

In second case, using setPartition restrict SQL query with given Partition
and a Key can be non-affinity key for this partition.
So, here task can be run on node that is primary for Key=1, but make a
remote query to node that is primary to partition=1.



On Sun, Oct 22, 2017 at 11:53 AM, dark <ek...@gmail.com> wrote:

> Result of the document
>
> setLocal is a local query
> setCollocated specifies that the data associated with Cache's Join function
> is local
>
> It seems like this.
> But why does the Partitioned Cache increase the GridQueryExecutor's Task on
> all nodes?
>
> Thank you.
>
>
>
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/
>



-- 
Best regards,
Andrey V. Mashenkov

Re: How to query to local partition cache data

[dark <ek...@gmail.com>]

Result of the document

setLocal is a local query
setCollocated specifies that the data associated with Cache's Join function
is local

It seems like this.
But why does the Partitioned Cache increase the GridQueryExecutor's Task on
all nodes?

Thank you.



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to query to local partition cache data

[dark <ek...@gmail.com>]

In addition, I am confused about exactly what each of the following setting
values that can be set in SqlFieldsQuery means.

setLocal
setCollocated

Can you explain the meaning of these by taking an example?



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/