[
https://issues.apache.org/jira/browse/HBASE-16586?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Matteo Bertozzi updated HBASE-16586:
------------------------------------
Attachment: HBASE-16586-v0.patch
v0 mostly add comments to explain the semantic of wait/wake. flip the
waitRegion() return value, add some asserts around and fixes a tryXLock where
we are already owner
> Procedure v2 - Cleanup sched wait/lock semantic
> -----------------------------------------------
>
> Key: HBASE-16586
> URL: https://issues.apache.org/jira/browse/HBASE-16586
> Project: HBase
> Issue Type: Sub-task
> Components: proc-v2
> Affects Versions: 2.0.0
> Reporter: Matteo Bertozzi
> Assignee: Matteo Bertozzi
> Fix For: 2.0.0
>
> Attachments: HBASE-16586-v0.patch
>
>
> For some reason waitEvent() and waitRegion() had a mismatching return value.
> unity the wait semantic in being: return true we wait, false we don't wait.
> procdures using hasLock = waitRegion() should change to hasLock =
> !waitRegion(). at the moment we have only DispatchMergingRegionsProcedure
> using it (in master).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
