[
https://issues.apache.org/jira/browse/HBASE-20685?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nihal Jain updated HBASE-20685:
-------------------------------
Description:
HBASE-11961 added docs for region state transitions. I have a few points in
mind:
* The RST graph says {{OFFLINE}} is a terminal state (i.e. regions of disabled
tables are set to {{OFFLINE}}), but with master branch disabling a table puts
the corresponding regions to {{CLOSED}} state.
* The {{CLOSED}} state has been marked as transient but as I understand it is
a terminal state with AMv2 design (unassigning a region puts it into {{CLOSED}}
state)
I assume there might be other discrepancies which I am not aware of. I think
this diagram needs to be re-looked and is out dated w.r.t. master.
was:
HBASE-11961 added docs for region state transitions. I have a few points in
mind:
* The RST graph says {{OFFLINE}} is a terminal state (i.e. regions of disabled
tables are set to OFFLINE), but with master branch disabling a table puts the
corresponding regions to {{CLOSED}} state.
* The CLOSED state has been marked as transient but as I understand it is a
terminal state with AMv2 design (unassigning a region puts it into CLOSED state)
I assume there might be other discrepancies which I am not aware of. I think
this diagram needs to be re-looked and is out dated w.r.t. master.
> Relook region state transition documentation
> --------------------------------------------
>
> Key: HBASE-20685
> URL: https://issues.apache.org/jira/browse/HBASE-20685
> Project: HBase
> Issue Type: Bug
> Components: documentation
> Reporter: Nihal Jain
> Priority: Major
>
> HBASE-11961 added docs for region state transitions. I have a few points in
> mind:
> * The RST graph says {{OFFLINE}} is a terminal state (i.e. regions of
> disabled tables are set to {{OFFLINE}}), but with master branch disabling a
> table puts the corresponding regions to {{CLOSED}} state.
> * The {{CLOSED}} state has been marked as transient but as I understand it
> is a terminal state with AMv2 design (unassigning a region puts it into
> {{CLOSED}} state)
> I assume there might be other discrepancies which I am not aware of. I think
> this diagram needs to be re-looked and is out dated w.r.t. master.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
