[ 
https://issues.apache.org/jira/browse/PHOENIX-1107?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14074501#comment-14074501
 ] 

Jesse Yates commented on PHOENIX-1107:
--------------------------------------

bq. First impression is that it looks like one big hack

Oh, is most certainly is a large hack - what gave it away, the unused column 
family or the overriding of the replication scopes?

bq. If the replication endpoint could use Phoenix APIs 

Having a separate phoenix endpoint just for doing the index updates seems 
overly complicated. Essentially, you would need to put in a custom replication 
codec that diverts the index updates and then writes them via a phoenix client 
again (which has multiple rounds trips, but over the WAN this time)... why not 
just turn on replication for the index table as well?

bq. What would a patch look like if we did replicate the index updates?

Just setting the CFs in the index table to include REPLICATION_SCOPE=GLOBAL and 
a single subclass of Replication to correctly handle the index updates as a 
WALActionListener (that method I posted above just needs to be overwritten). 
0.98 would probably be similar (not sure how the Replication code comes into 
play with the scopes, but I can't imagine its all that different).


> Support mutable indexes over replication
> ----------------------------------------
>
>                 Key: PHOENIX-1107
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-1107
>             Project: Phoenix
>          Issue Type: Bug
>    Affects Versions: 5.0.0, 3.1, 4.1
>            Reporter: Jesse Yates
>            Assignee: Jesse Yates
>         Attachments: phoenix-1107-3.0.v0
>
>
> Mutable indexes don't support usage with replication. For starters, the 
> replication WAL Listener checks the family of the edits, which can throw a 
> NPE for the IndexedKeyValue 



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Reply via email to