Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 24 Mar 2000 14:18:20 -0500
From:      "Troy Settle" <troy@picus.com>
To:        "Eric J. Schwertfeger" <ejs@bfd.com>
Cc:        <freebsd-questions@FreeBSD.ORG>
Subject:   RE: filesystem full?
Message-ID:  <BPEOKDIIGLACIBFDFLEPMEGKCAAA.troy@picus.com>
In-Reply-To: <Pine.BSF.4.10.10003241106100.85863-100000@harlie.bfd.com>

next in thread | previous in thread | raw e-mail | index | archive | help

Eric,

Thanks for this explanation.  It makes perfect sense.

I'm still going to grumble though... it's the nature of it all :)

-Troy



** -----Original Message-----
** From: Eric J. Schwertfeger [mailto:ejs@bfd.com]
** Sent: Friday, March 24, 2000 2:13 PM
** To: Troy Settle
** Cc: freebsd-questions@FreeBSD.ORG
** Subject: RE: filesystem full?
** 
** 
** On Fri, 24 Mar 2000, Troy Settle wrote:
** 
** > 
** > 
** > Hmm... bad taste to reply to one's own posts... but, I need to make a
** > correction:
** > 
** > 
** > ** If a process dies out, the kernel should release all 
** resources used by that
** > ** process.  The only processes using the /var filesystem, are the
** > ** default system
** > ** processes.  I can't make any claims otherwise.
** > 
** > After re-reading other replies, this should read:
** > 
** > If a process has a file open, the kernel should not allow a user 
** or another
** > process to delete said file.
** 
** According to unix filesystem semantics, deleting a file that is in use is
** perfectly legitimate, and changing this would in fact break software.
** 
** A file is a set of blocks on the disk pointed to by an inode.  A directory
** entry is a pointer to the inode.  When you open a file, the inode is
** resolved and the filename itself is no longer used.  The distinction
** becomes critical in cases like this, or when dealing with hard links,
** which are multiple directory entries for the same inode.
** 
** Now, when you delete or rename a file, you're not affecting the inode, but
** the directory entry.  However, the inode usage count gets decremented, and
** when it drops to zero, the inode and associated blocks are considered
** free.
** 
** Now, in the case where a program has a file open, the inode usage doesn't
** drop to 0 until the program closes the filehandle.
** 
** Am I being clear?  I've explained this to people that I work with, but
** email is different (no feedback, no gestures, etc).
** 
** It's a simple concept, but one that isn't expected, so most people trip
** over it their first time.
** 
** 


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?BPEOKDIIGLACIBFDFLEPMEGKCAAA.troy>