[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [rdiff-backup-users] Warning: Attempt to rename over same inode

From: Robert Nichols
Subject: Re: [rdiff-backup-users] Warning: Attempt to rename over same inode
Date: Fri, 04 Apr 2014 10:20:31 -0500
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:17.0) Gecko/20131030 Thunderbird/17.0.10

On 04/02/2014 05:49 AM, Dominic Raferd wrote:
On 29/03/2014 12:40, Marcus Schopen wrote:

I got these warnings today, when running an rdiff-backup on some of my
cyrus mailboxes. I did a lvm snap before running backup:

Warning: Attempt to rename over same
inode: /backup/var/spool/cyrus/mail/a/user/archive/nagios/rdiff-backup.tmp.863 
to /backup/var/spool/cyrus/mail/a/user/archive/nagios/515.

What is this about? A rdiff-backup -l /backup/ shows no problems,
rdiff-backup --verify /backup/ is fine (Every file verified


Marcus, see this old thread:
Andrew Ferguson's suggestion (he was the maintainer of rdiff-backup at the time)
is that it indicates an error in the underlying filesystem, but this is
unconfirmed. I presume that current versions of rdiff-backup include Andrew's
fix whereby the error is ignored by rdiff-backup which runs to completion.

It's not an error in the filesystem. It happens when the hard links among files
in the mirror somehow get out of sync with what is recorded in the rdiff-backup
metadata. I see it happen occasionally (yes, it's non-fatal now), but I've
never managed to generate a test case that will reproduce it. Running with the
"--no-compare-inode" option exacerbates the problems that rdiff-backup has with
hard links, and may be a factor for this one too. (I have a system with multiple
encrypted volumes that seem to get unlocked in a random order on each boot, but
I had to stop using the --no-compare-inode option because of the problems it
caused with hard links.)

Bob Nichols     "NOSPAM" is really part of my email address.
                Do NOT delete it.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]