commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bernd Eckenfels (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (VFS-545) DefaultFilesCache leaks closed filesystems
Date Sun, 16 Nov 2014 07:01:33 GMT

    [ https://issues.apache.org/jira/browse/VFS-545?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14208387#comment-14208387
] 

Bernd Eckenfels edited comment on VFS-545 at 11/16/14 7:00 AM:
---------------------------------------------------------------

My idea would be to simply use

{code}
    public void clear(FileSystem filesystem)
    {
        Map<FileName, FileObject> files = filesystemCache.remove(filesystem);
        if (files != null) files.clear(); // help GC
    }
{code}


was (Author: b.eckenfels):
My idea would be to simply use

{code}
    public void clear(FileSystem filesystem)
    {
        Map<FileName, FileObject> files = filesystemCache.remove(filesystem);
        files.clear(); // help GC
    }
{code}

> DefaultFilesCache leaks closed filesystems
> ------------------------------------------
>
>                 Key: VFS-545
>                 URL: https://issues.apache.org/jira/browse/VFS-545
>             Project: Commons VFS
>          Issue Type: Bug
>    Affects Versions: 2.0
>            Reporter: Bernd Eckenfels
>            Assignee: Bernd Eckenfels
>            Priority: Minor
>              Labels: leak
>             Fix For: 2.1
>
>
> The org.apache.commons.vfs2.cache.DefaultFilesCache does not remove the filesystem specific
cache if a filesystem gets cleared. This leads to the problem that instances if the FileSystem
kept alive after the FileProvider has closed the filesystem. This is not so much a problem
for a smaller number of filesystems configured for a prefix, but it is a problem for layered
or virtual filesystems which get created and destroyed. (See also VFS-544). The more advanced
filesystem caches support clearing the keys (but have other races I think).
> The behaviour is somewhat documented "lifetime of the FileSystemManager", but I don't
think it is expected or required.



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

Mime
View raw message