[
https://issues.apache.org/jira/browse/CASSANDRA-3272?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13116945#comment-13116945
]
Jackson Chung commented on CASSANDRA-3272:
------------------------------------------
"Why are you using EACH_QUORUM?"
-- no particular reason other than testing.
"I can't think of any reason someone would actually want to do EACH_QUORUM on a
read, when LOCAL_QUORUM would be sufficient if the write was done at
EACH_QUORUM. "
-- would it make sense to make such consistency level as an "invalid to use"
for READ then?
What if write (update, say add a new column/value to the key) is done via
LOCAL_QUORUM to DC1 but DC2 were down.
anti-entropy repair wasn't done
stop DC1 (for some reason this step is important)
now bring DC2 back up (assume HH is off)
read LQ from DC2..
you get an outdated result.
at least i think with EACH_QUORUM on a read, you should get an consistent
result if 2 nodes replicas agreed on each DC (in the example where RF=3 in each
DC)
(and you should get an UAE in the above example)
> READ Operation with CL=EACH_QUORUM succeed when a DC is down (RF=3)
> -------------------------------------------------------------------
>
> Key: CASSANDRA-3272
> URL: https://issues.apache.org/jira/browse/CASSANDRA-3272
> Project: Cassandra
> Issue Type: Bug
> Reporter: Jackson Chung
>
> "READ EACH_QUORUM: Returns the record with the most recent timestamp once
> a quorum of replicas in each data center of the cluster has responded."
> In other words, if a DC is down and the QUORUM could not be reached on that
> DC, read should fail.
> test case:
> - Cassandra version 0.8.6:
> INFO [main] 2011-09-28 22:26:24,297 StorageService.java (line 371) Cassandra
> version: 0.8.6
> - 6-node cluster with 2 DC and 3 node each. RF=3 in each DC:
> [default@Keyspace3] describe keyspace;
> Keyspace: Keyspace3:
> Replication Strategy: org.apache.cassandra.locator.NetworkTopologyStrategy
> Durable Writes: true
> Options: [DC2:3, DC1:3]
> Column Families:
> ColumnFamily: test
> Key Validation Class: org.apache.cassandra.db.marshal.BytesType
> Default column value validator: org.apache.cassandra.db.marshal.BytesType
> Columns sorted by: org.apache.cassandra.db.marshal.BytesType
> Row cache size / save period in seconds: 0.0/0
> Key cache size / save period in seconds: 200000.0/14400
> Memtable thresholds: 1.0875/1440/232 (millions of ops/minutes/MB)
> GC grace seconds: 864000
> Compaction min/max thresholds: 4/32
> Read repair chance: 1.0
> Replicate on write: true
> Built indexes: []
> all nodes are up, insert a row:
> $ nodetool -h localhost ring
> Address DC Rack Status State Load Owns Token
> 141784319550391026443072753096570088106
> 10.34.79.179 DC1 RAC1 Up Normal 11.13 KB 16.67% 0
> 10.34.70.163 DC2 RAC1 Up Normal 11.14 KB 16.67%
> 28356863910078205288614550619314017621
> 10.35.81.147 DC1 RAC1 Up Normal 11.14 KB 16.67%
> 56713727820156410577229101238628035242
> 10.84.233.170 DC2 RAC1 Up Normal 11.14 KB 16.67%
> 85070591730234615865843651857942052864
> 10.195.201.236 DC1 RAC1 Up Normal 11.14 KB 16.67%
> 113427455640312821154458202477256070485
> 10.118.147.73 DC2 RAC1 Up Normal 11.14 KB 16.67%
> 141784319550391026443072753096570088106
> - insert a value
> [default@Keyspace3] set
> test[utf8('test-key-1')][utf8('test-col')]=utf8('test-value');
> Value inserted.
> sanity check (cli connects to a node in DC1) :
> [default@Keyspace3] consistencylevel as EACH_QUORUM;
>
> Consistency level is set to 'EACH_QUORUM'.
> [default@Keyspace3] get test[utf8('test-key-1')];
> => (column=746573742d636f6c, value=test-value, timestamp=1317249361722000)
> Returned 1 results
> shut down DC2:
> $ nodetool -h localhost ring
> Address DC Rack Status State Load Owns
> Token
>
> 141784319550391026443072753096570088106
> 10.34.79.179 DC1 RAC1 Up Normal 51.86 KB 16.67%
> 0
> 10.34.70.163 DC2 RAC1 Down Normal 51.88 KB 16.67%
> 28356863910078205288614550619314017621
> 10.35.81.147 DC1 RAC1 Up Normal 47.5 KB 16.67%
> 56713727820156410577229101238628035242
> 10.84.233.170 DC2 RAC1 Down Normal 51.88 KB 16.67%
> 85070591730234615865843651857942052864
> 10.195.201.236 DC1 RAC1 Up Normal 47.5 KB 16.67%
> 113427455640312821154458202477256070485
> 10.118.147.73 DC2 RAC1 Down Normal 51.88 KB 16.67%
> 141784319550391026443072753096570088106
> [default@Keyspace3] get test[utf8('test-key-1')];
> => (column=746573742d636f6c, value=746573742d76616c7565,
> timestamp=1317249361722000)
> Returned 1 results.
> tried with pycassaShell:
> >>> col_fam.get('test-key-1',read_consistency_level=pycassa.ConsistencyLevel.EACH_QUORUM)
> OrderedDict([('test-col', 'test-value')])
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
