[
https://issues.apache.org/jira/browse/ZOOKEEPER-1879?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13897764#comment-13897764
]
Flavio Junqueira commented on ZOOKEEPER-1879:
---------------------------------------------
My comment in ZOOKEEPER-1573 was more along the lines of checking the concrete
case of a node not existing because it is deleted later in the txn log and the
snapshot has already incorporated it. I think we can have such a check inlined.
A consistency checker would be nice so that we can do overall checks, but I see
the goal of a consistency being different than the one of the inline check I'm
mentioning.
> improve the correctness checking of txn log replay
> --------------------------------------------------
>
> Key: ZOOKEEPER-1879
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1879
> Project: ZooKeeper
> Issue Type: Improvement
> Components: server
> Affects Versions: 3.4.6, 3.5.0
> Reporter: Patrick Hunt
> Fix For: 3.4.7, 3.5.0
>
>
> In ZOOKEEPER-1573 we decided to fix an issue by relaxing some of the
> checking. Specifically when the sequence of txns is as follows:
> * zxid 1: create /prefix/a
> * zxid 2: create /prefix/a/b
> * zxid 3: delete /prefix/a/b
> * zxid 4: delete /prefix/a
> the log may fail to replay.
> We addressed this by relaxing a check, which is essentially invalid for this
> case, but is important in finding corruptions of the datastore.
> We should add this check back with proper validation of correctness.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
