phoenix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <>
Subject [jira] [Commented] (PHOENIX-3534) Support multi region SYSTEM.CATALOG table
Date Wed, 04 Jul 2018 23:20:04 GMT


ASF GitHub Bot commented on PHOENIX-3534:

Github user JamesRTaylor commented on a diff in the pull request:
    --- Diff: phoenix-core/src/it/java/org/apache/phoenix/end2end/ ---
    @@ -372,6 +378,31 @@ public void testViewAndTableAndDrop() throws Exception {
             // drop table cascade should succeed
             conn.createStatement().execute("DROP TABLE " + fullTableName + " CASCADE");
    +        validateViewDoesNotExist(conn, fullViewName1);
    +        validateViewDoesNotExist(conn, fullViewName2);
    +    }
    +    @Test
    +    public void testRecreateDroppedTableWithChildViews() throws Exception {
    --- End diff --
    These new tests are good. These are testing that the left over metadata doesn't impact
the re-creation of a table since we don't make the RPC to delete views when a base table is
dropped, right? Do you think there'd be any issues if part of the rows for a view were there
(i.e. say that the create view failed, but some of the rows were written)? Might be good to
have a test like this - you could set it up by using HBase APIs to manually delete some rows
of a view.

> Support multi region SYSTEM.CATALOG table
> -----------------------------------------
>                 Key: PHOENIX-3534
>                 URL:
>             Project: Phoenix
>          Issue Type: Bug
>            Reporter: James Taylor
>            Assignee: Thomas D'Silva
>            Priority: Major
>             Fix For: 5.0.0, 4.15.0
>         Attachments: PHOENIX-3534.patch
> Currently Phoenix requires that the SYSTEM.CATALOG table is single region based on the
server-side row locks being held for operations that impact a table and all of it's views.
For example, adding/removing a column from a base table pushes this change to all views.
> As an alternative to making the SYSTEM.CATALOG transactional (PHOENIX-2431), when a new
table is created we can do a lazy cleanup  of any rows that may be left over from a failed
DDL call (kudos to [~lhofhansl] for coming up with this idea). To implement this efficiently,
we'd need to also do PHOENIX-2051 so that we can efficiently find derived views.
> The implementation would rely on an optimistic concurrency model based on checking our
sequence numbers for each table/view before/after updating. Each table/view row would be individually
locked for their change (metadata for a view or table cannot span regions due to our split
policy), with the sequence number being incremented under lock and then returned to the client.

This message was sent by Atlassian JIRA

View raw message