cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alex Petrov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-12011) Logback shutdown hook races with StorageServiceShutdownHook
Date Tue, 04 Oct 2016 07:26:20 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-12011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15544609#comment-15544609
] 

Alex Petrov commented on CASSANDRA-12011:
-----------------------------------------

This issue has been fixed in [CASSANDRA-12509]. Logging is the last thing that shuts down
now, after everything else.

> Logback shutdown hook races with StorageServiceShutdownHook
> -----------------------------------------------------------
>
>                 Key: CASSANDRA-12011
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12011
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Joel Knighton
>            Assignee: Alex Petrov
>            Priority: Minor
>             Fix For: 2.2.x, 3.0.x, 3.x
>
>
> In {{StorageService}}, we add a JVM shutdown hook called {{StorageServiceShutdownHook}}
that shuts down gossip and flushes writes for non durable keyspaces, among other things. 
> In {{logback.xml}}, we add a JVM shutdown hook that tears down logback resources.
> Since JVM shutdown hooks are started concurrently, the logback shutdown hook almost always
completes during the {{StorageServiceShutdownHook}}, which means that any log statements in
this hook will not work since the logback resources have been torn down.
> This makes debugging the {{StorageServiceShutdownHook}} challenging. We should do our
own logback teardown that does not run until after the completion of any Cassandra-oriented
JVM shutdown hooks such as the {{StorageServiceShutdownHook}}.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message