Control: tag -1 +unreproducible

Hi Paul

Just for the record: so far I could not reproduce this bug.  On our productive 
system there are enough changes every day that there is always a diff.

In view of the upcoming jessie freeze I looked into the matter during the last 
couple of weeks.  I set up a test system to try out a few things and could not 
reproduce your situation.

> On 15 Jun 2014, at 05:35, Paul Wise <p...@debian.org> wrote:
> 
> Package: ldap-git-backup
> Version: 1.0.7-1
> Severity: normal
> 
> Every night I get a mail like this. Looks like it is caused by
> ldap-git-backup running git commit even if there is nothing to commit,
> in which case git commit returns failure (1) instead of success (0).
> 
> Subject: Cron <pabs@chianamo> /usr/sbin/ldap-git-backup --ldif-cmd '...' 
> --backup-dir ... --commit-msg "Import $(date --rfc-3339=seconds)"
> 
> commit --message=Import 2014-06-14 00:00:10+08:00 --date=1402675211: command 
> returned error: 1

In a vanilla installation with slapd and its default data two consecutive 
manual runs of ldap-git-backup both give an successful exit code of 0 and no 
output.  The second call is a no-op.  The git log just shows the first backup.  
This is how it is meant to behave.

If I manually run a git commit in the backup directory without any pending 
changes in the git index I do get a non-successful exit code.  However, 
ldap-git-backup does not propagate this up the call chain in my setup.

A vague guess is that some git config setting could trigger the behaviour you 
are seeing.  But unless I can reproduce a test case I hesitate to change my 
code.  Chances are I might introduce some subtle bugs that would prevent making 
a backup in some cases.

I agree that if the situation arises as it does in your case that this is a 
bug.  I also think the severity is appropriately set.

Regards     — Elmar

--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to