[
https://issues.apache.org/jira/browse/RATIS-585?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17753321#comment-17753321
]
Tsz-wo Sze commented on RATIS-585:
----------------------------------
Once the original log entry is committed, the Raft algorithm guarantees that
the log entry remains committed even after server restarts. The logMetadata is
only for the server to learn about its commit index. It the logMetadata is
lost, the server still will learn about its commit index later on. So, this
is not a problem.
> LogMetadata does not wait for the metada operations to finish before
> acknowleding watch requests
> ------------------------------------------------------------------------------------------------
>
> Key: RATIS-585
> URL: https://issues.apache.org/jira/browse/RATIS-585
> Project: Ratis
> Issue Type: Bug
> Components: server
> Affects Versions: 0.4.0
> Reporter: Mukul Kumar Singh
> Priority: Major
>
> LogMetadata call on receiving a quorum does not wait for the file IO to
> complete before acknowledging the watch requests. This can lead to problems
> on restart where a watch request acknowledged back to the client is not
> persisted.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
