Bug#41427: marked as done (Odd failure to fsync() importanttmp)
Your message dated Wed, 21 Jul 1999 20:52:06 +0100 (BST)
with message-id <email@example.com>
and subject line Bug#41427: Odd failure to fsync() importanttmp
has caused the attached bug report to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what I'm
talking about this indicates a serious mail system misconfiguration
somewhere. Please contact me immediately.)
(administrator, Debian bugs database)
Received: (at submit) by bugs.debian.org; 16 Jul 1999 21:44:37 +0000
Received: (qmail 18468 invoked from network); 16 Jul 1999 21:44:37 -0000
Received: from mimis.host4u.net (@188.8.131.52)
by master.debian.org with SMTP; 16 Jul 1999 21:44:37 -0000
Received: from hildy (1Cust134.tnt2.lancaster.pa.da.uu.net [184.108.40.206])
by mimis.host4u.net (8.8.5/8.8.5) with ESMTP id QAA19048
for <firstname.lastname@example.org>; Fri, 16 Jul 1999 16:44:28 -0500
Received: from theorb by hildy with local (Exim 3.02 #1 (Debian))
for <email@example.com>; Fri, 16 Jul 1999 17:12:03 -0400
Date: Fri, 16 Jul 1999 17:11:59 -0400
From: "James M. Mastros" <firstname.lastname@example.org>
Subject: Odd failure to fsync() importanttmp
Message-ID: <[🔎] 19990716171149.A32667@hildy>
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.6i
Here's an odd failure mode for you guys...
After a bad shutdown (hardware problem: the unlabled green button is reset,
not turbo), dpkg refused to install any packages. Specificly, the
fsync(fileno(importanttmp)) on line 234 of lib/dbmodify.c filed with EIO.
Under investigation, I found that the EIO had to have come from the ext2
layer, and is therefor not dpkg's fault (as any problem with the input
should have been caught in the vfs layer).
So, the bug is just that dpkg checks the return value of fsync at all -- or
at least it being fatal. The package will install just fine without the
file makeing it to disk (it's a temp file anyway), so it's the test's fault
that dpkg broke for me (by a kind of warped reasoning; it's acatualy an ext2
problem, but one that's probably close to imposible to diagnose.)
(Yes, I did verify that "#if 0"ing out those two lines allows packages to be
installed correctly. And filed a false bug against autoconf in the process,
to boot, but that's another story, and will be told another time.)
-=- James Mastros
To err is human, to be pandatic is oft just plain annoying.
(See my recent posting to debian-devel.)