File Component readLock strategy

[Craig Tataryn <cr...@tataryn.net>]

Has anyone had any luck using a readLock strategy other than "changed"?  I've tried them all, and "changed" seems to be the only one that works consistently when I'm copying a large file into a directory the File Component is monitoring.

All other's allow the File Component to execute before the file is completely copied into the directory.

$mvn -X --version
Apache Maven 3.0-beta-3 (r990787; 2010-08-30 07:44:03-0500)
Java version: 1.6.0_20
Java home: /System/Library/Frameworks/JavaVM.framework/Versions/1.6.0/Home
Default locale: en_US, platform encoding: MacRoman
OS name: "mac os x" version: "10.6.4" arch: "x86_64" Family: "mac"

Craig.

--
Craig Tataryn
site: http://www.basementcoders.com/
podcast: http://www.basementcoders.com/?feed=podcast
itunes: http://itunes.apple.com/podcast/the-basement-coders
irc: ThaDon on freenode #basementcoders, ##wicket, #papernapkin
twitter: craiger

Re: File Component readLock strategy

[Claus Ibsen <cl...@gmail.com>]

Hi

Its actually a bit tricky to detect if a file is locked by another
process or not.
Although we have the java.nio API it seems to not work consistent
across OS's. And I actually think its stated in the docs somewhere,
that this can be the case.

Also note that NIO locks is JVM wide, which means from an unit test
you cannot lock a file by one thread, and from the same JVM see the
file lock.
And if you forget to unlock the NIO lock, that file is locked until
you restart the JVM. Can be a caveat for big app servers, if you have
a rouge app holding a lock.

Then you can try to obtain the lock on the file by trying to rename it
before reading, if you can rename then you assume the other party is
not currently writing to it.
But I think some OS and apps may allow the OS to rename a file which
currently is being written by that app.

That leaves you with checking the file modification and timestamp and
wait until its "steady".

But NIO 2 should have a file watcher API. I assume it may be better at
providing notifications when a file has been finished writing.
So we can wait for Java 7 :)



On Thu, Oct 7, 2010 at 1:07 AM, Craig Tataryn <cr...@tataryn.net> wrote:
> Has anyone had any luck using a readLock strategy other than "changed"?  I've tried them all, and "changed" seems to be the only one that works consistently when I'm copying a large file into a directory the File Component is monitoring.
>
> All other's allow the File Component to execute before the file is completely copied into the directory.
>
> $mvn -X --version
> Apache Maven 3.0-beta-3 (r990787; 2010-08-30 07:44:03-0500)
> Java version: 1.6.0_20
> Java home: /System/Library/Frameworks/JavaVM.framework/Versions/1.6.0/Home
> Default locale: en_US, platform encoding: MacRoman
> OS name: "mac os x" version: "10.6.4" arch: "x86_64" Family: "mac"
>
> Craig.
>
> --
> Craig Tataryn
> site: http://www.basementcoders.com/
> podcast: http://www.basementcoders.com/?feed=podcast
> itunes: http://itunes.apple.com/podcast/the-basement-coders
> irc: ThaDon on freenode #basementcoders, ##wicket, #papernapkin
> twitter: craiger
>
>



-- 
Claus Ibsen
Apache Camel Committer

Author of Camel in Action: http://www.manning.com/ibsen/
Open Source Integration: http://fusesource.com
Blog: http://davsclaus.blogspot.com/
Twitter: http://twitter.com/davsclaus