[
https://issues.apache.org/jira/browse/HBASE-13937?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14592814#comment-14592814
]
Duo Zhang commented on HBASE-13937:
-----------------------------------
For isServerReachable, I think we can say a server is 'dead' if one of the
following conditions is satisfied
1. Server tells us it is dead(I'm not sure whether a
RegionServerStoppedException is enough, maybe it will be thrown before
regionserver completely shutdown?)
2. It is a server with another start code.
3. We get a connection refused(not connect timeout).
So I think remove the code is reasonable, but we should catch a connection
refused exception then(Does our rpc framework throw this exception out, and
also we do not need retry here...)? Otherwise if we do not restart a
regionserver then we will be stuck in the loop for a long time...
And also I do not think it is safe to return 'not reachable' if timeout...
Thanks.
> Partially revert HBASE-13172
> -----------------------------
>
> Key: HBASE-13937
> URL: https://issues.apache.org/jira/browse/HBASE-13937
> Project: HBase
> Issue Type: Sub-task
> Components: Region Assignment
> Reporter: Enis Soztutar
> Assignee: Enis Soztutar
> Fix For: 0.98.14, 1.2.0, 1.1.1, 1.3.0
>
> Attachments: hbase-13937_v1.patch
>
>
> HBASE-13172 is supposed to fix a UT issue, but causes other problems that
> parent jira (HBASE-13605) is attempting to fix.
> However, HBASE-13605 patch v4 uncovers at least 2 different issues which are,
> to put it mildly, major design flaws in AM / RS.
> Regardless of 13605, the issue with 13172 is that we catch
> {{ServerNotRunningYetException}} from {{isServerReachable()}} and return
> false, which then puts the Server to the {{RegionStates.deadServers}} list.
> Once it is in that list, we can still assign and unassign regions to the RS
> after it has started (because regular assignment does not check whether the
> server is in {{RegionStates.deadServers}}. However, after the first assign
> and unassign, we cannot assign the region again since then the check for the
> lastServer will think that the server is dead.
> It turns out that a proper patch for 13605 is very hard without fixing rest
> of broken AM assumptions (see HBASE-13605, HBASE-13877 and HBASE-13895 for a
> colorful history). For 1.1.1, I think we should just revert parts of
> HBASE-13172 for now.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
