incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Watanabe Maki <>
Subject Re: snapshot files locked
Date Thu, 15 Mar 2012 00:06:49 GMT



On 2012/03/15, at 9:04, Watanabe Maki <> wrote:

> It's true on unix, but you can't delete hard links of opened files on windows.
> Try following:
> 1  Create a text file ORIG.TXT
>  Now you have hard link LINK.TXT .
> 3  Open ORIG.TXT with MS Word.
>    It returns error.
> Interesting thing is that you can delete link.txt if you open orig.txt with notepad.

> Notepad may close file after load, or it may be using different API.
> On 2012/03/15, at 5:04, Jim Newsham <> wrote:
>> Hi Maki,
>> Thanks for the reply.  Yes, I understand that snapshots are hard links.  However,
my understanding is that removing any hard-linked files just removes the link (decrementing
the link counter of the file on disk) -- it does not delete the file itself nor remove any
other links which may be pointing at the file.  To confirm my understanding, I tested this
in Windows by terminating Cassandra and then deleting all files in the snapshot dir.  None
of the corresponding files in the parent keyspace directory were removed.
>> Regards,
>> Jim
>> On 3/13/2012 9:29 PM, Maki Watanabe wrote:
>>> snapshot files are "hardlink"s of the original sstables.
>>> As you know, on windows, you can't delete files opened by other process.
>>> If you try to delete the "hardlink", windows thinks you try to delete
>>> the sstables in production.
>>> maki
>>> 2012/3/14 Jim Newsham<>:
>>>> Hi,
>>>> I'm using Cassandra 1.0.8, on Windows 7.  When I take a snapshot of the
>>>> database, I find that I am unable to delete the snapshot directory (i.e.,
>>>> dir named "{datadir}\{keyspacename}\snapshots\{snapshottag}") while
>>>> Cassandra is running:  "The action can't be completed because the folder
>>>> a file in it is open in another program.  Close the folder or file and try
>>>> again".  If I terminate Cassandra, then I can delete the directory with no
>>>> problem.  Is there a reason why Cassandra must hold onto these files?
>>>> Thanks,
>>>> Jim

View raw message