phoenix-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Samarth Jain (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (PHOENIX-3970) Ensure that automatic partial index rebuilds are served from the index handler pool
Date Thu, 22 Jun 2017 07:13:00 GMT

    [ https://issues.apache.org/jira/browse/PHOENIX-3970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16058880#comment-16058880
] 

Samarth Jain commented on PHOENIX-3970:
---------------------------------------

The chances of running into a deadlock when using the same handler pool for data table and
index table writes is higher. 

Let's take the simplest case where you have two region servers with the data table regions
and the index table regions on both of them. If index handler pool on RS1 is being used to
write to data table, then the index mutations originating from RS2 to RS1 could be blocked
because all the index handlers on RS1 are taken. Similarly RPCs originating from RS1 to RS2
would be blocked because all the index handlers on RS2 are not able to make any progress resulting
in a deadlock.

> Ensure that automatic partial index rebuilds are served from the index handler pool
> -----------------------------------------------------------------------------------
>
>                 Key: PHOENIX-3970
>                 URL: https://issues.apache.org/jira/browse/PHOENIX-3970
>             Project: Phoenix
>          Issue Type: Bug
>            Reporter: Lars Hofhansl
>            Assignee: Lars Hofhansl
>         Attachments: 3970.txt, 3970-v2.txt
>
>
> This (and other issues) have rendered multiple larger cluster inoperable.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message