[
https://issues.apache.org/jira/browse/HDFS-1974?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13081951#comment-13081951
]
Aaron T. Myers commented on HDFS-1974:
--------------------------------------
Thanks a lot for addressing my concerns, Suresh. The latest patch looks a lot
better to me.
bq. My concern with the common change is adding a notion of StandbyException to
IPC. We could have a separate discussion about whether the notion of
StandbyException belongs to IPC layer or to the service. One thing I have been
thinking is for upper layers to install error/exception handlers. That upper
layers could use exception handler to handle StandbyException.
My point isn't that StandbyException is necessarily the right approach, just
that I do think the IPC layer should be responsible for the mechanics of client
failover, and thus that there should be some way for an exception thrown by a
remote service to indicate to the client that a request should be retried
against another server. But, this discussion should probably happen on a
separate JIRA.
bq. I also think that having a method to transition to every state, means too
many methods in HAState. So I have changed the patch to use setState() and
setStateInternal(). See if it makes more sense now.
I agree. The latest patch looks much better to me in this regard.
Two final comments. +1 once these are addressed:
# The two-argument constructor in {{UnsupportedActionException}} doesn't
actually use the {{action}} parameter.
# I don't think {{listCorruptFileBlocks}} is a write operation.
> HA: Introduce active and standby states to the namenode
> -------------------------------------------------------
>
> Key: HDFS-1974
> URL: https://issues.apache.org/jira/browse/HDFS-1974
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: name-node
> Reporter: Suresh Srinivas
> Assignee: Suresh Srinivas
> Attachments: HDFS-1974.1.patch, HDFS-1974.2.patch, HDFS-1974.3.patch,
> HDFS-1974.4.patch, HDFS-1974.5.patch, HDFS-1974.patch
>
>
> Currently namenode supports active, secondary and backup roles. To support
> namenode high availability, active and standby states are needed. Note that
> this is different from the existing notion of namenode role, where a namenode
> cannot transition from one role to the other.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
