[ https://issues.apache.org/jira/browse/CASSANDRA-1012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12860019#action_12860019 ]
Erick Tryzelaar commented on CASSANDRA-1012: -------------------------------------------- So here's what I'm seeing in my log files with my inserts and gets. And by saying "inserting into machine1", I mean that my client is connecting to machine1, but machine1, machine2, and machine3 are all in a ring together and appear to have normal communication with each other: % nodetool -h machine1 ring Address Status Load Range Ring mGdY41r8RNGYLXsF 10.0.0.1 Up 20.23 KB A7fozEmmpyDXlE7e |<--| 10.0.0.3 Up 5.44 KB fQKDOQPHr2x8n1zG | | 10.0.0.2 Up 5.44 KB mGdY41r8RNGYLXsF |-->| % nodetool -h machine1 streams Mode: Normal Not sending any streams. Not receiving any streams. % nodetool -h machine2 streams Mode: Normal Not sending any streams. Not receiving any streams. % nodetool -h machine3 streams Mode: Normal Not sending any streams. Not receiving any streams. ------ inserting into machine1: machine1: DEBUG [ROW-READ-STAGE:4] 2010-04-22 14:10:43,513 StorageProxy.java (line 757) weakreadlocal reading SliceByNamesReadCommand(table='Keyspace1', key=3131, columnParent='QueryPath(columnFamilyName='Standard1', superColumnName='null', columnName='null')', columns=[6a6f626964,]) DEBUG [pool-1-thread-2] 2010-04-22 14:10:43,788 CassandraServer.java (line 417) batch_mutate DEBUG [pool-1-thread-2] 2010-04-22 14:10:43,869 StorageProxy.java (line 299) insert writing local key [...@163b4b1e machine2: (nothing in log) machine3: (nothing in log) ------ getting against machine1: machine1: (nothing in log) machine2: DEBUG [pool-1-thread-4] 2010-04-22 14:07:22,825 CassandraServer.java (line 232) get_slice DEBUG [ROW-READ-STAGE:7] 2010-04-22 14:07:22,826 StorageProxy.java (line 757) weakreadlocal reading SliceByNamesReadCommand(table='Keyspace1', key=31, columnParent='QueryPath(columnFamilyName='Standard1', superColumnName='null', columnName='null')', columns=[6a6f626964,]) machine3: (nothing in log) ------ getting against machine2: machine1: DEBUG [ROW-READ-STAGE:4] 2010-04-22 14:03:10,038 ReadVerbHandler.java (line 94) Read key [...@3dd06d02; sending response to 153934@/10.0.0.2 machine2: DEBUG [Timer-0] 2010-04-22 14:03:03,747 LoadDisseminator.java (line 37) Disseminating load info ... DEBUG [pool-1-thread-8] 2010-04-22 14:03:10,012 CassandraServer.java (line 232) get_slice DEBUG [pool-1-thread-8] 2010-04-22 14:03:10,013 StorageProxy.java (line 341) weakreadremote reading SliceByNamesReadCommand(table='Keyspace1', key=31, columnParent='QueryPath(columnFamilyName='Standard1', superColumnName='null', columnName='null')', columns=[6a6f626964,]) DEBUG [pool-1-thread-8] 2010-04-22 14:03:10,013 StorageProxy.java (line 352) weakreadremote reading SliceByNamesReadCommand(table='Keyspace1', key=31, columnParent='QueryPath(columnFamilyName='Standard1', superColumnName='null', columnName='null')', columns=[6a6f626964,]) from 153934@/10.0.0.1 DEBUG [RESPONSE-STAGE:8] 2010-04-22 14:03:10,024 ResponseVerbHandler.java (line 44) Processing response on an async result from 153934@/10.0.0.1 machine3: (nothing in log) > Problems reading remote rows in trunk? > -------------------------------------- > > Key: CASSANDRA-1012 > URL: https://issues.apache.org/jira/browse/CASSANDRA-1012 > Project: Cassandra > Issue Type: Bug > Components: Core > Affects Versions: 0.7 > Reporter: Erick Tryzelaar > > I've run into a problem with my small 3 node cluster where quorum reads do > not return anything. I'm inserting 10 rows into a fresh database with > replication factor 3. I've inserted with QUORUM, but when I try to read back > at QUORUM no data gets returned. However, if I use the consistency level ONE, > I do get results. > I've dug a bit into this problem, and it appears that internally > ThriftServer.getSlice is finding the column family for the given key, but > it's not finding any of the columns, so thrift doesn't return anything. > In order to confirm there being a problem, I dropped the RF down to 1 and > rebuilt the cluster. I then reinserted some data, and I can only read data > off of one machine. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.