xfs
[Top] [All Lists]

Re: Might have found a bug...

To: Russell Cattelan <cattelan@xxxxxxxxxxx>
Subject: Re: Might have found a bug...
From: patl@xxxxxxxxxxxxxxx (Patrick J. LoPresti)
Date: 12 May 2001 11:51:14 -0400
Cc: linux-xfs@xxxxxxxxxxx
Distribution: local
In-reply-to: <mit.lcs.mail.linux-xfs/3AFC1625.44E539C8@xxxxxxxxxxx>
References: <200105111625.f4BGP9u10062@xxxxxxxxxxx> <a3AFC1625.44E539C8@xxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Gnus/5.0808 (Gnus v5.8.8) Emacs/20.7
Russell Cattelan <cattelan@xxxxxxxxxxx> writes:

> XFS uses delayed allocation / delayed write by default any data
> written immediacy before a crash probably won't be on disk, this is
> known behavior and isn't considered a bug, it's a trade off.

Is this true even if I call fsync() or fdatasync() on the file?

I am wondering about this because some applications (e.g., qmail)
create a temp file, call fsync() to flush it to stable storage, use
rename() to "commit" the file to its final home.

With XFS, is this a reliable way to update a file on disk, or not?

 - Pat

<Prev in Thread] Current Thread [Next in Thread>