[jira] [Comment Edited] (HBASE-21863) narrow down the double-assignment race window

["Sergey Shelukhin (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-21863?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16765553#comment-16765553 ] 

Sergey Shelukhin edited comment on HBASE-21863 at 2/12/19 12:05 AM:
--------------------------------------------------------------------

[~stack] can you elaborate on extra states from deadline? If the message did expire (master is no longer waiting), we avoid doing something master doesn't expect. If it doesn't expire and we respond with error, it happens before any work, so the master will just handle it like a regular error. It's not ideal but should be rare and doesn't add new states.

I'd like to add it to region report, however it causes some issues: HBASE-21522 and especially HBASE-21531 (that is resolved as a dup of HBASE-21421, without fixing the actual race) that is a race that happens a lot.
So it was removed in HBASE-21421.
I filed a separate JIRA to add it back. I think given that TRSP is one place that sort of knows what's going on, it's a good place to have it for now :) Ignoring some RS reporting region open doesn't seem to be correct. 
I can replace with a more specific exception.

There's discussion in the other bug about the root cause...
However, for production use it's better to prevent double assignment due to unknown bugs, to avoid data loss...


was (Author: sershe):
[~stack] can you elaborate on extra states from deadline? If the message did expire (master is no longer waiting), we avoid doing something master doesn't expect. If it doesn't expire and we respond with error, it happens before any work, so the master will just handle it like a regular error. It's not ideal but should be rare and doesn't add new states.

I'd like to add it to region report, however it causes some issues: HBASE-21522 and especially HBASE-21531 that is a race that happens a lot.
So it was removed in HBASE-21421.
I filed a separate JIRA to add it back. I think given that TRSP is one place that sort of knows what's going on, it's a good place to have it for now :) Ignoring some RS reporting region open doesn't seem to be correct. 
I can replace with a more specific exception.

There's discussion in the other bug about the root cause...
However, for production use it's better to prevent double assignment due to unknown bugs, to avoid data loss...

> narrow down the double-assignment race window
> ---------------------------------------------
>
>                 Key: HBASE-21863
>                 URL: https://issues.apache.org/jira/browse/HBASE-21863
>             Project: HBase
>          Issue Type: Bug
>            Reporter: Sergey Shelukhin
>            Assignee: Sergey Shelukhin
>            Priority: Major
>         Attachments: HBASE-21863.patch
>
>
> See HBASE-21862.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)