[
https://issues.apache.org/jira/browse/FLUME-2298?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13874654#comment-13874654
]
Ashish Paliwal commented on FLUME-2298:
---------------------------------------
Some minor comments
1. "If DC_foo is unable to find a slave it will continue to work as a normal
channel until a slave becomes available"
IMHO, the behavior should be configurable. It's should be User's choice. User
may assume that by using DC, they have HA for channel, whereas in this case
this may not be true.
2. "Death of a Master" Section - Can use Zookeeper state information to detect
dead nodes, by using watches.
3. How shall the Event replication to Slave channel work? Shall we use some
replication mechanism or create a Slave source and use slave sink at master to
write to slaves. This part of DC design is not very clear to me.
For last comment, I don't have details of discussion, as it did not happen on
ML, but IMHO, Slaves are also Channels and would exhibit same properties as a
normal channel. Additional state information of message processing is not
needed.
> Distributed Channel
> -------------------
>
> Key: FLUME-2298
> URL: https://issues.apache.org/jira/browse/FLUME-2298
> Project: Flume
> Issue Type: New Feature
> Components: Channel
> Reporter: Ted Malaska
> Assignee: Ted Malaska
> Attachments: FlumeDistributedChannelDesign.0.1.pdf
>
>
> This channel will allow for events to be persisted with a plugable method on
> more then one agent or node.
> The goal is to gain the following benefits:
> 1. Events will continue to flow to sinks with out loss or with out large
> delay even in the case of node failure.
> 2. Protect against loss in the case of a complete single node failure
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
