[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [rdiff-backup-users] rdiff-backup-regress script query - need to be
From: |
Dominic Raferd |
Subject: |
Re: [rdiff-backup-users] rdiff-backup-regress script query - need to be 'nobody' or 'root' |
Date: |
Fri, 23 Jun 2017 14:26:17 +0100 |
On 23 June 2017 at 12:53, Ron Leach <address@hidden> wrote:
> List, good morning,
>
> Having messed up the latest backup increment of our quite-large filesystem
> and, having long been a very happy user of Dominic's rdiff-backup-regress
> script, I thought I'd better regress the messed-up increment.
>
> On running the regress script as user 'ron', the script advised that I
> needed to be user 'nobody' or 'root'. But, after su-ing to root, the
> script warned me:
>
> address@hidden:/home/ron# ./rdiff-backup-regress.sh -n 1
> /mnt/backserver/Backups
>
> rdiff-backup-regress.sh v1.0 [25 Aug 2016] by Dominic (-h for help)
> =======================
>
> You are user 'root', not 'nobody', which may result in changed ownership
> of some files.
> Are you sure you wish to continue (y/-)? n
> Exiting, no changes made
> address@hidden:/home/ron#
>
> I've not had this warning before. Many of the files in this backup are
> used by MS Windows clients, and managed by samba; I think this may be where
> the 'nobody' user is coming from. The files are also served over NFS to
> linux clients, perhaps they are marked as 'nobody'.
>
> I think I ought not mess up the file ownerships, so I wanted to take heed
> of the script's warning. I don't know how to become user 'nobody' - I
> don't think there is a 'nobody' password.
>
> Before I dig a deeper hole, has someone else bumped into this when using
> rdiff-backup-regress script and (if so) what did they find was the best way
> forward? I probably do need to regress this increment, really.
>
> Grateful for any comment,
Hi Ron, I'm glad my script has helped you in the past. The warning is from
my script, not from rdiff-backup and was based on my experience. The safest
way to run the script IMO is as user nobody i.e. the same user as created
the archive, you might have to create a password temporarily for this user
and then remove it afterwards. If you run it as root there is a risk that
some files in the repository are left readable and/or writeable only by
root so future rdiff-backup sessions by the original user could fail. If
you have to run the regression as root I advise taking a backup of the
entire repository beforehand. Unless anyone knows different?