[
https://issues.apache.org/jira/browse/ASTERIXDB-2467?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16662966#comment-16662966
]
ASF subversion and git services commented on ASTERIXDB-2467:
------------------------------------------------------------
Commit 776e1c25b11b85aa752671ff8992dcc509e28913 in asterixdb's branch
refs/heads/master from luochen
[ https://git-wip-us.apache.org/repos/asf?p=asterixdb.git;h=776e1c2 ]
[ASTERIXDB-2467][STO] Fix locking protocol with bad tuples
- user model changes: no
- storage format changes: no
- interface changes: no
Details:
- The current locking protocol is not correct when there are bad tuples,
e.g., duplicates. When locking fails, the write must push partial frames
so that the previous acquired locks are released. However, if there are
bad tuples in the frame, after bad tuples are removed from the frame, the
state of pushed partial frames will be completely lost.
Change-Id: I86baa1f21ef7390777c8e2b0ce226095528e7f18
Reviewed-on: https://asterix-gerrit.ics.uci.edu/3003
Reviewed-by: Xikui Wang <[email protected]>
Sonar-Qube: Jenkins <[email protected]>
Tested-by: Jenkins <[email protected]>
Integration-Tests: Jenkins <[email protected]>
Reviewed-by: Murtadha Hubail <[email protected]>
> Concurrency control protocol is not correct when there are bad tuples
> ---------------------------------------------------------------------
>
> Key: ASTERIXDB-2467
> URL: https://issues.apache.org/jira/browse/ASTERIXDB-2467
> Project: Apache AsterixDB
> Issue Type: Bug
> Components: STO - Storage
> Reporter: Chen Luo
> Priority: Major
>
> The current deadlock-free locking protocol pushes partial frames when the try
> locking fails. However, when there are bad tuples (e.g., duplicates) in the
> frame, the state of partial frames is always reset after the bad tuple is
> removed from frame.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
