curator-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Shevek (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CURATOR-42) Background guaranteed delete considers NoNode to be a failed delete, and retries it
Date Mon, 24 Feb 2014 17:30:23 GMT

    [ https://issues.apache.org/jira/browse/CURATOR-42?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13910545#comment-13910545
] 

Shevek commented on CURATOR-42:
-------------------------------

It may be hard to deterministically produce the necessary situation. I'll think about it.
At the time, I was playing with locking and election, so I assume the background delete in
question was a lock removal. However, it may be sufficient to just background-guaranteed-delete
a nonexistent node. I'm trying to remember if it checks for existence before starting the
loop, in which case it's necessary to deterministically produce the race in test.

FYI, the worst race I ever hit in ZK: The server told me it had failed to create a ZNode,
but it had succeeded.

> Background guaranteed delete considers NoNode to be a failed delete, and retries it
> -----------------------------------------------------------------------------------
>
>                 Key: CURATOR-42
>                 URL: https://issues.apache.org/jira/browse/CURATOR-42
>             Project: Apache Curator
>          Issue Type: Bug
>            Reporter: Shevek
>            Assignee: Jordan Zimmerman
>             Fix For: awaiting-response
>
>
> In delete(), NoNode should be a special case which succeeds instantly, and does not consider
the delete failed.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message