This doesn’t actually guarantee the behavior you think it does. There’s no actual way to guarantee this behavior in Cassandra, as far as I can tell. A long time ago there was a ticket for a “coordinator only” consistency level, which is nearly trivial to implement, but the use case is so narrow that it’s unlikely to ever be done.
Here’s an example trace on system_auth, where all nodes are replicas (RF=N),
and the data is fully repaired (data exists on the local node). The coordinator
STILL chooses a replica other than itself (far more likely to see this behavior
on a keyspace with a HIGH replication factor, this particular cluster is N in
the hundreds):
cqlsh> tracing on;
Tracing is already enabled. Use TRACING OFF to disable.
cqlsh> CONSISTENCY local_one;
Consistency level set to LOCAL_ONE.
cqlsh> select name from system_auth.users where name='jjirsa' limit 1;
name
--------
jjirsa
(1 rows)
Tracing session: 5ffdee70-12d5-11e6-ad58-317180027532
activity
| timestamp | source | source_elapsed
-------------------------------------------------------------------------------------------------+----------------------------+--------------+----------------
Execute CQL3 query | 2016-05-05 15:23:52.919000 | x.y.z.150 | 0
Parsing select * from system_auth.users where name='jjirsa' limit 1;
[SharedPool-Worker-7] | 2016-05-05 15:23:52.919000 | x.y.z.150 | 100
Preparing statement
[SharedPool-Worker-7] | 2016-05-05 15:23:52.920000 | x.y.z.150 | 194
reading data from /x.y.z.151
[SharedPool-Worker-7] | 2016-05-05 15:23:52.920000 | x.y.z.150 | 965
Sending READ message to /x.y.z.151
[MessagingService-Outgoing-/x.y.z.151] | 2016-05-05 15:23:52.921000 | x.y.z.150
| 1072
REQUEST_RESPONSE message received from /x.y.z.151
[MessagingService-Incoming-/x.y.z.151] | 2016-05-05 15:23:52.924000 | x.y.z.150
| 5433
Processing response from /x.y.z.151
[SharedPool-Worker-4] | 2016-05-05 15:23:52.924000 | x.y.z.150 | 5595
READ message received from /x.y.z.150
[MessagingService-Incoming-/x.y.z.150] | 2016-05-05 15:23:52.927000 | x.y.z.151
| 104
Executing single-partition query on users
[SharedPool-Worker-6] | 2016-05-05 15:23:52.928000 | x.y.z.151 | 2251
Acquiring sstable references
[SharedPool-Worker-6] | 2016-05-05 15:23:52.929000 | x.y.z.151 | 2353
Merging memtable tombstones
[SharedPool-Worker-6] | 2016-05-05 15:23:52.929000 | x.y.z.151 | 2414
Partition index with 0 entries found for sstable 384
[SharedPool-Worker-6] | 2016-05-05 15:23:52.929000 | x.y.z.151 | 2829
Seeking to partition beginning in data file
[SharedPool-Worker-6] | 2016-05-05 15:23:52.930000 | x.y.z.151 | 2913
Skipped 0/1 non-slice-intersecting sstables, included 0 due to tombstones
[SharedPool-Worker-6] | 2016-05-05 15:23:52.930000 | x.y.z.151 | 3263
Merging data from memtables and 1 sstables
[SharedPool-Worker-6] | 2016-05-05 15:23:52.931000 | x.y.z.151 | 3289
Read 1 live and 0 tombstone cells
[SharedPool-Worker-6] | 2016-05-05 15:23:52.931000 | x.y.z.151 | 3323
Enqueuing response to /x.y.z.150
[SharedPool-Worker-6] | 2016-05-05 15:23:52.931000 | x.y.z.151 | 3411
Sending REQUEST_RESPONSE message to /x.y.z.150
[MessagingService-Outgoing-/x.y.z.150] | 2016-05-05 15:23:52.932000 | x.y.z.151
| 3577
Request complete | 2016-05-05 15:23:52.924649 | x.y.z.150 | 5649
From: Varun Barala
Reply-To: "user@cassandra.apache.org"
Date: Thursday, May 5, 2016 at 2:40 AM
To: "user@cassandra.apache.org"
Subject: Re: Read data from specific node in cassandra
Hi Siddharth Verma,
You can define consistency level LOCAL_ONE.
and you can applyh consistency level during statement creation.
like this -> statement.setConsistencyLevel(ConsistencyLevel.LOCAL_ONE);
On Thu, May 5, 2016 at 3:55 PM, Siddharth Verma <verma.siddha...@snapdeal.com>
wrote:
Hi,
We have a 3 node cluster in DC1, where replication factor of keyspace is 3.
How can i read data only from one particular node in java driver?
Thanks,
Siddharth Verma
smime.p7s
Description: S/MIME cryptographic signature
