[ 
https://issues.apache.org/jira/browse/CASSANDRA-1853?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12972620#action_12972620
 ] 

Jonathan Ellis commented on CASSANDRA-1853:
-------------------------------------------

shouldn't applyModels be the One True Place to call CFMetaData.apply?  I'd like 
to r/m the apply calls in the constructor and in CassandraServer to avoid 
confusion.

> changing row cache save interval is reflected in 'describe keyspace' on node 
> it was submitted to, but not nodes it was propagated to
> ------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-1853
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-1853
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Peter Schuller
>            Assignee: Gary Dusbabek
>            Priority: Minor
>             Fix For: 0.7.0
>
>         Attachments: v1-0001-apply-CF-metadata-updates-on-replicas.txt
>
>
> This is minor unless it indicates a bigger issue. On our test cluster 
> (running cassandra 0.7 branch from today) we noticed that on submission of a 
> new row cache save period, such as:
>    update column family KeyValue with row_cache_save_period=3600;
> The change would be reflected in describe_keyspace() (describe keyspace ... 
> in cassandra-cli) on the node to which the schema migration was submitted, 
> but not on other nodes in the cluster.
> This in spite of the schema migration having propagated, judging by 
> Schema['Last Migration'] being identical on all nodes. It is not n and of 
> itself is not a big problem, but it does give the impression that the 
> migrations have trouble propagating throughout the cluster even though they 
> do.
> (I had a quick (only) look in the code paths of migration application and did 
> not find any obvious special casing of the node that happens to be local. 
> Filing bug instead, hoping someone knows off hand what the reason is.)

-- 
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