[
https://issues.apache.org/jira/browse/PHOENIX-583?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Taylor resolved PHOENIX-583.
----------------------------------
Resolution: Fixed
Fix Version/s: 5.0.0
4.0.0
3.0.0
> Base tenant table PK must lead with tenant_id and tenant_type_id columns
> ------------------------------------------------------------------------
>
> Key: PHOENIX-583
> URL: https://issues.apache.org/jira/browse/PHOENIX-583
> Project: Phoenix
> Issue Type: Task
> Affects Versions: 3.0-Release
> Reporter: elevine
> Assignee: elevine
> Fix For: 3.0.0, 4.0.0, 5.0.0
>
>
> From @jamestaylor:
> One overall comment: we should formalize the link between the tenant-specific
> table name and the key prefix in the data row. Otherwise, often times a query
> will run over all "virtual" tables instead of just the data for the specific
> one. My proposal would be to:
> * in addition the "base/parent table property" supplied at DDL time, require
> an additional "type id" property. It's possible, when we get support for
> sequences, that the server could generate this, but for now we could require
> it. This "type id" (i.e. the key prefix) would get cached in PTable so that
> the client will always have this value.
> * data rows would be expected to conform to the following format:
> * when you get to the issue about enforcing tenant_id visibility (i.e. adding
> tenant-id to the where clause), you'd also automatically add the key prefix
> to the where clause. You'd always be able to get the type-id, as it would be
> cached in the PTable which was looked up by name.
> Related to tenant-specific table feature (issues -356 and -296).
--
This message was sent by Atlassian JIRA
(v6.2#6252)
