[45] in The Cryptographic File System users list
Re: Possible Directory Bug in 1.4 Beta 2
daemon@ATHENA.MIT.EDU (Rob Stampfli)
Thu Apr 23 01:46:04 1998
From owner-cfs-users@research.att.com Thu Apr 23 05:46:04 1998
Return-Path: <owner-cfs-users@research.att.com>
Delivered-To: cfs-mtg@bloom-picayune.mit.edu
Received: (qmail 3252 invoked from network); 23 Apr 1998 05:46:03 -0000
Received: from unknown (HELO rumor.research.att.com) (192.20.225.9)
by bloom-picayune.mit.edu with SMTP; 23 Apr 1998 05:46:03 -0000
Received: from research.att.com ([135.207.30.100]) by rumor; Thu Apr 23 01:41:39 EDT 1998
Received: from amontillado.research.att.com ([135.207.24.32]) by research-clone; Thu Apr 23 01:44:33 EDT 1998
Received: from nsa.research.att.com (majordomo@nsa.research.att.com [135.207.24.155])
by amontillado.research.att.com (8.8.7/8.8.7) with ESMTP id BAA22311;
Thu, 23 Apr 1998 01:44:26 -0400 (EDT)
Received: (from majordomo@localhost) by nsa.research.att.com (8.7.3/8.7.3) id BAA06243 for cfs-users-list; Thu, 23 Apr 1998 01:41:01 -0400 (EDT)
X-Authentication-Warning: nsa.research.att.com: majordomo set sender to owner-cfs-users@nsa.research.att.com using -f
Received: from research.att.com (research-clone.research.att.com [135.207.30.100]) by nsa.research.att.com (8.7.3/8.7.3) with SMTP id BAA06239 for <cfs-users@nsa.research.att.com>; Thu, 23 Apr 1998 01:40:59 -0400 (EDT)
Received: from elektro.cmhnet.org ([192.188.133.3]) by research-clone; Thu Apr 23 01:41:33 EDT 1998
Received: from colnet by elektro.cmhnet.org with uucp
(Smail3.1.29.1 #1) id m0ySEka-0000ZQC; Thu, 23 Apr 98 01:40 EDT
Received: from kd8wk.cmhnet.org by colnet.cmhnet.org with smtp
(Smail3.1.28.1 #4) id m0ySEbs-0008F9C; Thu, 23 Apr 98 01:31 EDT
Received: by kd8wk.cmhnet.org (Smail3.1.28.1 #4)
id m0ySEba-0000nqC; Thu, 23 Apr 98 01:31 EDT
Message-Id: <m0ySEba-0000nqC@kd8wk.cmhnet.org>
Date: Thu, 23 Apr 98 01:31 EDT
From: res@kd8wk.cmhnet.org (Rob Stampfli)
To: "Alex Raftis" <alex@echidna.doverpacific.com>
Cc: cfs-users@research.att.com
Subject: Re: Possible Directory Bug in 1.4 Beta 2
Sender: owner-cfs-users@research.att.com
Precedence: bulk
In (not too) recent email Alex Raftis writes:
>
>I recently noticed that as I was moving files into a crypted file system, the
>files were "disappearing". I say "disappearing" because they don't appear in
>a directory listing. However, they do exist. Ie, if I happen to remember the
>name, I can open and view the file, even though it's not appearing the in the
>directory listing. Looking into the problem futher, I find that the raw disk
>has 2187 (4293 including dot) files in the directory, but the clear text
>directory only shows a 121 files. Obviously, something is a little weird
>here.
>
>Unfortuneately, I can't use cname to try and list the files since I'm using
>the latest version of the file structure.
>
>Has anyone else encountered such a problem? Is there someway to get my files
>to show up again in plain text?
>
>BTW, I'm using cfs under a OPENSTEP 4.2 (BSD 4.3 filesystem) port of the
>daemon.
Since I have seen no other responses, let me suggest a few things:
First, you have an awful lot of files in one directory. I have never
created that large a directory to my knowledge, and frankly, I have never
noticed the "lost file" phenomenon on my Suns. (When I read your email,
I went back and audited a few directories just to convince myself I wasn't
missing something.) That said, I did have some problems getting cfs to work
over nfs on an HP workstation. While playing around with the rsize and
wsize parameters on the cfs mount in an attempt to resolve these problems,
I discovered that when I set these parameters to small values (something
like 256 or below), an "ls" of the /crypt directory stopped producing any
filenames, although, like you, I could operate on the files contained
therein if I knew the name. At the time I attributed it to an nfs glitch,
but it does have the feel of what you are experiencing. Check your mount
parms or play with them to see if they have any effect.
Second, does a reboot of cfsd fix the problem temporarily? I have seen
some squirrelly things on cfsds which have been running for a while.
Nothing like what you describe, but glitchy things like cattaches which
appear to work, but in reality don't mount anything. In fact, some time
ago I started killing and restarting cfsd from cron every night. For me,
cfs runs a lot more robustly this way, not to mention this keeps it from
becoming such a memory hog. And, if I accidently forget and go away
leaving a /crypt directory mounted this effectively does a cdetach on
everything.
Question: do small directories work OK? If so, how big do they have to
grow before you can no longer do an "ls" on them?
Hope this helps. I really enjoy having and using cfs. BTW, anyone know
what has happened to Matt?
--
Rob Stampfli rob@colnet.cmhnet.org The Bill of Rights: It was a
614-864-9377 HAM RADIO: kd8wk@w8cqk.oh good thing while it lasted...
