[ 
http://jira.amdatu.org/jira/browse/AMDATUCASSANDRA-36?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=11169#comment-11169
 ] 

Ivo Ladage - van Doorn commented on AMDATUCASSANDRA-36:
-------------------------------------------------------

Added a Unit test that could reproduce the issue mentioned in 
AMDATUCASSANDRA-65. 
In fact I added two unit tests; one using the deprecated Thrift PM and one 
using the new Hector PM.
Now the issue can still be reproduced using the Thrift PM, but cannot be 
reproduced using the Hector PM.
So apparently, we did something wrong in our implementation of the Thrift based 
PM. Note that we did find more related issues in the Thrift based PM, like 
tombstone records still being returned as real rows (see AMDATUCASSANDRA-15).

So the final conclusion on this issue is: the problem was caused by an 
incorrect implementation of our Thrift-based PM. Switching to the Hector API 
resolved the problem. Case closed.
I will commit the unit tests, but disable the Thrift PM one (as that will fail).

> Investigate Cassandra consistency problem with asynchronous write operations
> ----------------------------------------------------------------------------
>
>                 Key: AMDATUCASSANDRA-36
>                 URL: http://jira.amdatu.org/jira/browse/AMDATUCASSANDRA-36
>             Project: Amdatu Cassandra
>          Issue Type: Bug
>          Components: Persistence Manager
>            Reporter: Ivo Ladage - van Doorn
>             Fix For: 0.2.0
>
>
> As described on 
> http://www.amdatu.org/confluence/pages/viewpage.action?pageId=1605717
> it seems that it is not possible to ensure that this always works in a 
> single-node cassandra cluster:
> (pseudo code)
> - setValue(key, value)
> - AssertTrue(getValue(key)==value)
> It needs to be investigated if this is really true.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
_______________________________________________
Amdatu-developers mailing list
[email protected]
http://lists.amdatu.org/mailman/listinfo/amdatu-developers

Reply via email to