[
https://issues.apache.org/jira/browse/HADOOP-8247?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13248724#comment-13248724
]
Aaron T. Myers commented on HADOOP-8247:
----------------------------------------
{quote}
Add a new flag dfs.ha.automatic-failover.enabled, which is set per-nameservice
or globally
Add a new RequestInfo structure as a parameter to all the HAServiceProtocol
methods. This currently just has one field, which indicates what type of client
the request is on behalf of. It can either be a user (manual CLI failover),
ZKFC (auto failover), or USER_FORCE – indicating that it's a user who wants to
avoid this safety check.
In the NN, if auto-failover is enabled, disallow HA requests from users. If
it's not enabled, disallow HA requests from ZKFCs.
In the ZKFC, disallow startup if auto-failover is disabled
{quote}
All this makes a lot of sense to me, Todd. The only question I have is whether
or not it really makes sense to add a RequestInfo structure, instead of just an
extra parameter whose value is defined by a simple 3-element enum. What else do
you envision being added to the RequestInfo structure?
> Auto-HA: add a config to enable auto-HA, which disables manual FC
> -----------------------------------------------------------------
>
> Key: HADOOP-8247
> URL: https://issues.apache.org/jira/browse/HADOOP-8247
> Project: Hadoop Common
> Issue Type: Improvement
> Components: auto-failover, ha
> Affects Versions: Auto Failover (HDFS-3042)
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Attachments: hadoop-8247.txt
>
>
> Currently, if automatic failover is set up and running, and the user uses the
> "haadmin -failover" command, he or she can end up putting the system in an
> inconsistent state, where the state in ZK disagrees with the actual state of
> the world. To fix this, we should add a config flag which is used to enable
> auto-HA. When this flag is set, we should disallow use of the haadmin command
> to initiate failovers. We should refuse to run ZKFCs when the flag is not
> set. Of course, this flag should be scoped by nameservice.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
