[
https://issues.apache.org/jira/browse/HBASE-18609?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16164917#comment-16164917
]
Mike Drob commented on HBASE-18609:
-----------------------------------
Would it make sense to have ClusterStatus return some kind of error code for
statuses that it didn't fetch?
Visually inspecting the code, I can see this kind of error occurring in client
logic:
{code}
ClusterStatus status = getClusterStatus(EnumSet.of(Option.LIVE_SERVERS,
Option.BACKUP_MASTERS));
ServerName master = status.getBackupMasters();
{code}
Actually, this example wouldn't compile because of return types, but if we're
the point is if we're switching what we get and ask for the wrong thing it will
probably fail silently and in strange ways. May be better to explicitly throw a
StatusNotAvailableException or something of the like?
> Apply ClusterStatus#getClusterStatus(EnumSet<Option>) in code base
> ------------------------------------------------------------------
>
> Key: HBASE-18609
> URL: https://issues.apache.org/jira/browse/HBASE-18609
> Project: HBase
> Issue Type: Improvement
> Reporter: Chia-Ping Tsai
> Assignee: Reid Chan
> Fix For: 2.0.0
>
> Attachments: HBASE-18609.master.001.patch,
> HBASE-18609.master.002.patch
>
>
> HBASE-15511 enable us to get the cluster status by scope, and after
> refactoring in HBASE-18621. We should apply it in code base so as to prevent
> the useless information.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
