[
https://issues.apache.org/jira/browse/HBASE-18448?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16105975#comment-16105975
]
Ajay Jadhav commented on HBASE-18448:
-------------------------------------
[~ram_krish]:
1. In case we don't find the region in the first run due to on going split,
we'll get the regions in next iteration. Yes, I do see the pain that customers
have to face but we are looking at solutions to overcome this manual effort and
warm up the replica automatically.
The reason why we have kept this API invocation manual is due to performance
impact and the frequency of use. The use case that we were trying to solve
didn't involve heavy use of this API due to which we have kept it manual for
now. But yes, in case of write/ update heavy workloads, this solution is not at
all scalable.
2. I looked at the read replica support that HBase already had but in that
case, the primary has an idea about the replicas. So, it can send updates
across. In this case, the replica itself is a separate cluster and primary
cluster doesn't even know about the existence of replicas.
[~anoop.hbase] I haven't looked at the CP endpoints yet but this sounds like a
good idea. I'll take a look at CP and update the patch.
My only concern here is that we are already exposing this as an API.
> Added support for refreshing HFiles through API and shell
> ---------------------------------------------------------
>
> Key: HBASE-18448
> URL: https://issues.apache.org/jira/browse/HBASE-18448
> Project: HBase
> Issue Type: Improvement
> Affects Versions: 2.0.0, 1.3.1
> Reporter: Ajay Jadhav
> Assignee: Ajay Jadhav
> Priority: Minor
> Fix For: 1.4.0
>
> Attachments: HBASE-18448.branch-1.001.patch,
> HBASE-18448.branch-1.002.patch
>
>
> In the case where multiple HBase clusters are sharing a common rootDir, even
> after flushing the data from
> one cluster doesn't mean that other clusters (replicas) will automatically
> pick the new HFile. Through this patch,
> we are exposing the refresh HFiles API which when issued from a replica will
> update the in-memory file handle list
> with the newly added file.
> This allows replicas to be consistent with the data written through the
> primary cluster.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
