[ 
http://jira.amdatu.org/jira/browse/AMDATUCASSANDRA-138?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Ivo Ladage - van Doorn resolved AMDATUCASSANDRA-138.
----------------------------------------------------

    Resolution: Fixed

Added support for concurrent schema modifications. The internal Cassandra 
client in the CassandraDaemonService is replaced by a Thrift client to the 
first node in the lexicographic sorted list of IP addresses of live nodes. As a 
result all schema modifications on all nodes in a cluster are send to the same 
node (to which we refer as the schema master). 
                
> Support concurrent schema modifications in a Cassandra cluster
> --------------------------------------------------------------
>
>                 Key: AMDATUCASSANDRA-138
>                 URL: http://jira.amdatu.org/jira/browse/AMDATUCASSANDRA-138
>             Project: Amdatu Cassandra
>          Issue Type: Improvement
>          Components: Cassandra daemon
>    Affects Versions: 0.2.1
>            Reporter: Ivo Ladage - van Doorn
>            Assignee: Ivo Ladage - van Doorn
>             Fix For: 0.2.2
>
>
> Currently, concurrent schema modifications in a Cassandra cluster may cause 
> ConfigurationExceptions. For example, if a new tenant is deployed to all 
> nodes in the cluster, all nodes will try (concurrently) to add a keyspace and 
> ColumnFamily's to that new keyspace. This will cause ConfigurationExceptions, 
> as Cassandra assumes that schema modifications are always executed on a 
> single node in a single thread.
> To resolve this issue, all schema modifications should be handles by the 
> CassandraDaemonService, which ensures the schema modifications to to executed 
> by a single thread by making its methods synchronized. This still fails in a 
> cluster however. To resolve this issue:
> - Schema modifications should not be executed by the local Cassandra daemon, 
> but by one 'elected' node of the cluster
> - Which node will execute scheme modifications is determined by ordering all 
> live nodes in the cluster (i.e. by IP address) and picking the first one

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
http://jira.amdatu.org/jira/secure/ContactAdministrators!default.jspa
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