[
https://issues.apache.org/jira/browse/MESOS-1039?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13912181#comment-13912181
]
Benjamin Hindman commented on MESOS-1039:
-----------------------------------------
Let's make sure that we get 1040 done before doing this so that we don't
introduce weird read semantics.
> Log coordinator does not need to wait for the local replica to learn the
> value for a write
> ------------------------------------------------------------------------------------------
>
> Key: MESOS-1039
> URL: https://issues.apache.org/jira/browse/MESOS-1039
> Project: Mesos
> Issue Type: Improvement
> Reporter: Jie Yu
>
> Once a quorum of OK responses are received, the log coordinator can just
> declare the write operation to be successful, without waiting for the local
> replica to learn that value.
> However, this will complex the read semantics here because the leading
> replica no longer has up-to-date information. For example, the following code
> may have an exception being thrown:
> Log.Writer writer(log);
> Log.Position p = writer.append(...);
> Log.Reader reader(log);
> List<Log.Entry> entries = reader.read(p, p);
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
