[
https://issues.apache.org/jira/browse/ZOOKEEPER-4925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17948346#comment-17948346
]
Kezhu Wang commented on ZOOKEEPER-4925:
---------------------------------------
Added a test case for this:
https://github.com/kezhuw/zookeeper/commit/c9e00b87bd0c952d382250d7b911896f52e92407.
Only 3.9.3 is affected
> Diff sync introduce hole in stale follower's committedLog which cause data
> loss in leading
> ------------------------------------------------------------------------------------------
>
> Key: ZOOKEEPER-4925
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-4925
> Project: ZooKeeper
> Issue Type: Bug
> Components: server
> Affects Versions: 3.9.3
> Reporter: Kezhu Wang
> Assignee: Kezhu Wang
> Priority: Critical
>
> There are two variants of {{ZooKeeperServer::processTxn}}. Those two variants
> diverge in behavior since ZOOKEEPER-3484. {{processTxn(Request request)}}
> pops outstanding change from {{outstandingChanges}} and adds txn to
> {{committedLog}} for follower to sync in addition to what
> {{processTxn(TxnHeader hdr, Record txn)}} does. The {{Learner}} uses
> {{processTxn(TxnHeader hdr, Record txn)}} to commit txn to memory after
> ZOOKEEPER-4394, which means it leaves {{committedLog}} untouched in
> {{SYNCHRONIZATION}} phase.
> In above case, a stale follower will have hole in its {{committedLog}} after
> joining cluster. The stale follower will propagate the in memory hole to
> other stale nodes after becoming leader. This causes data loss.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
