[
https://issues.apache.org/jira/browse/HBASE-19339?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chia-Ping Tsai updated HBASE-19339:
-----------------------------------
Attachment: HBASE-19339.v0.patch
The root cause is that our flush suppose the size of snapshot won't be changed
after requiring the update lock from region, but the eager policy still run the
in-memory compaction which can change the flushable size.
Instead of using the size got from {{getFlushable}}, we should honor the size
from snapshot since the size got from snapshot is stable for all policies.
> Eager policy results in the negative size of memstore
> -----------------------------------------------------
>
> Key: HBASE-19339
> URL: https://issues.apache.org/jira/browse/HBASE-19339
> Project: HBase
> Issue Type: Bug
> Reporter: Chia-Ping Tsai
> Assignee: Chia-Ping Tsai
> Priority: Critical
> Fix For: 2.0.0-beta-1
>
> Attachments: HBASE-19339.v0.patch
>
>
> It is a follow-up of HBASE-19266.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
