[
https://issues.apache.org/jira/browse/HBASE-19002?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16220748#comment-16220748
]
Josh Elser commented on HBASE-19002:
------------------------------------
bq. This hook has store passed to it (compared to postFlush() above this). I
wonder if a different counter than flushCounter can be used to track the
occurrences.
Admittedly, this example feels a little contrived. There are two flush
code-paths: one is "flush regions on this RS to get the memstore utilization
down" and the other is "flush this specific region". IMO, they're still both
"flushes" happening, but they might not really be considered equivalent (e.g.
the one flush would just be creating a single new file, whereas the other may
create multiple).
bq. How about using the same type as that of the cell ?
So, I was struggling to figure out how to convert the {{byte}} which {{Cell}}
provides back into a {{DataType}} value. However, then I realized I should
really just make this only work on Puts which solves the problem.
Will update shortly.
> Introduce more examples to show how to intercept normal region operations
> -------------------------------------------------------------------------
>
> Key: HBASE-19002
> URL: https://issues.apache.org/jira/browse/HBASE-19002
> Project: HBase
> Issue Type: Sub-task
> Components: Coprocessors
> Reporter: Duo Zhang
> Assignee: Josh Elser
> Priority: Minor
> Fix For: 2.0.0-alpha-4
>
> Attachments: HBASE-19002.001.branch-2.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
