[ 
https://issues.apache.org/jira/browse/CASSANDRA-1623?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12922237#action_12922237
 ] 

Jason Tanner commented on CASSANDRA-1623:
-----------------------------------------

Patch tested on my install and it resolved my issue.

> BUG: secondaryIndexes AND multiple index expressions can cause timesouts 
> -------------------------------------------------------------------------
>
>                 Key: CASSANDRA-1623
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1623
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.7 beta 2
>         Environment: centos 5.5
>            Reporter: Jason Tanner
>            Assignee: Jonathan Ellis
>             Fix For: 0.7.0
>
>         Attachments: 1623.txt
>
>
> 1. Given this Column Family definition
>     Column Family Name: Requests
>       Column Family Type: Standard
>       Column Sorted By: org.apache.cassandra.db.marshal.UTF8Type
>       Column Metadata:
>         Column Name: requested
>           Validation Class: org.apache.cassandra.db.marshal.UTF8Type
>           Index Type: KEYS
>         Column Name: requestor
>           Validation Class: org.apache.cassandra.db.marshal.UTF8Type
>           Index Type: KEYS
> If I have an entry that has the following column/value pairs:
> "request-uuid1" : [  { "requested","person-uuid1" }, { 
> "requestor","person-uuid2"}, { "is_confirmed","true" } ]
> If I do an index lookup (pseudo coded) :
> get_index_slices( Connection,
>                                  ColumnParent.column_family="Requests",
>                                  [ { "requested","eq", "person-uuid1" }, { 
> "is_confirmed","eq", "false" } ],      % Index Expressions
>                                  "",100,   % StartKey, KeyCount
>                                  "","",false,100   % StartCol, EndCol, 
> Reversed, ColCount )
> for "requested" = "person-uuid1" and "is_confirmed" = false 
> then I get the following entries in my log and the request times out along 
> with all other requests on all clients.
> DEBUG [pool-1-thread-10] 2010-10-15 19:00:27,878 CassandraServer.java (line 
> 531) scan
> DEBUG [pool-1-thread-10] 2010-10-15 19:00:27,897 StorageProxy.java (line 563) 
> restricted single token match for query [0,0]
> DEBUG [pool-1-thread-10] 2010-10-15 19:00:27,897 StorageProxy.java (line 649) 
> scan ranges are [0,0]
> DEBUG [pool-1-thread-10] 2010-10-15 19:00:27,925 StorageProxy.java (line 669) 
> reading org.apache.cassandra.db.indexscancomm...@42a6eb from 
> 5...@localhost/127.0.0.1
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,931 SliceQueryFilter.java (line 121) 
> collecting 0 of 1: null:false:0...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,933 SliceQueryFilter.java (line 121) 
> collecting 0 of 2147483647: is_confirmed:false:4...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,934 SliceQueryFilter.java (line 121) 
> collecting 1 of 2147483647: request_type:false:6...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,935 SliceQueryFilter.java (line 121) 
> collecting 2 of 2147483647: requested:false:5...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,935 SliceQueryFilter.java (line 121) 
> collecting 3 of 2147483647: requested_network:false:5...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,936 SliceQueryFilter.java (line 121) 
> collecting 4 of 2147483647: requestor:false:5...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,937 SliceQueryFilter.java (line 121) 
> collecting 5 of 2147483647: requestor_network:false:5...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,942 SliceQueryFilter.java (line 121) 
> collecting 0 of 1: null:false:0...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,943 SliceQueryFilter.java (line 121) 
> collecting 0 of 1: null:false:0...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,945 SliceQueryFilter.java (line 121) 
> collecting 0 of 1: null:false:0...@1287103291
> DEBUG [ReadStage:2] 2010-10-15 19:00:27,946 SliceQueryFilter.java (line 121) 
> collecting 0 of 1: null:false:0...@1287103291
>  this last line repeats forever until I stop the server.
> If instead I do the lookup where both terms match or just the last term 
> matches then nothing goes wrong, I get a valid (empty or otherwise) result 
> set.
> It only seems to happen if the 2nd expression does not match.
> I am using the very latest code from trunk.
> Jason
>                                        

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to