phoenix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chinmay Kulkarni (Jira)" <j...@apache.org>
Subject [jira] [Created] (PHOENIX-5496) Ensure that we handle all server-side mutation codes on the client
Date Fri, 27 Sep 2019 00:27:00 GMT
Chinmay Kulkarni created PHOENIX-5496:
-----------------------------------------

             Summary: Ensure that we handle all server-side mutation codes on the client
                 Key: PHOENIX-5496
                 URL: https://issues.apache.org/jira/browse/PHOENIX-5496
             Project: Phoenix
          Issue Type: Bug
    Affects Versions: 4.15.0, 5.1.0
            Reporter: Chinmay Kulkarni
             Fix For: 4.15.1, 5.1.1


There are many instances throughout wherein we set a certain error mutation code in the RPC
callback, however we do not handle these mutation codes on the client.

For example: 

If the metadata rows for a tableKey are no longer in that SYSCAT region, checkTableKeyInRegion()
fails, the metadata for this table is not written to SYSCAT and [the TABLE_NOT_IN_REGION
mutation code is set|https://github.com/apache/phoenix/blob/11997d48d1957cf613526f01c5ccbe2812cf095d/phoenix-core/src/main/java/org/apache/phoenix/coprocessor/MetaDataEndpointImpl.java#L1785-L1790].

This is handled for 1 retry inside [CQSI.metaDataCoprocessorExec|https://github.com/apache/phoenix/blob/11997d48d1957cf613526f01c5ccbe2812cf095d/phoenix-core/src/main/java/org/apache/phoenix/query/ConnectionQueryServicesImpl.java#L1568-L1570],
but if this happens again, it is returned back to the client where it goes to the default
case and succeeds.

Apart from the fact that partial metadata updates are possible leading to orphan metadata
rows in system tables, this also wrongly returns success for clients even though there is
no record of that table/view being created inside Phoenix's system tables.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Mime
View raw message