Re: 3.0 on Debian Wheezy: setup-kolab failing because of not running 389

On Wednesday, September 26, 2012 16:57:18 Jeroen van Meeuwen wrote:
> On Tuesday, September 25, 2012 09:37:09 PM Johannes Graumann wrote:
> > Jeroen van Meeuwen wrote:
> > > On Tuesday, September 25, 2012 04:07:19 PM Johannes Graumann wrote:
> > >> So I keep digging into this and have been exploring what the
> > >> responsible
> > >> 
> > >> > /usr/lib/python2.7/dist-packages/pykolab/setup/setup_ldap.py
> > >> 
> > >> is actually doing.
> > >> 
> > >> First off I find that "/usr/sbin/setup-ds-admin.pl" seems to be called
> > >> "/usr/sbin/setup-ds-admin" here, so
> > >> 
> > >> > 225c225
> > >> > <     setup_ds_admin = "/usr/sbin/setup-ds-admin.pl"
> > >> > ---
> > >> > 
> > >> > >     setup_ds_admin = "/usr/sbin/setup-ds-admin"
> > >> 
> > >> might be in order.
> > > 
> > > It's actually the '-admin' part of 'setup-ds-admin' that causes the
> > > process to require the semaphore to be created.
> > > 
> > > Using just 'setup-ds' does succeed for me.
> > 
> > The semaphore issue is solved (see other part of this thread) by making
> > sure /dev/shm is actually mounted as tmpfs ... the error described above
> > is independent of the semaphore isse.
> 
> Right - that may resolve the issue for your deployment but it will still
> fail on a standard Debian installation (permission errors), which we can't
> sensibly change through packaging or application fixes.
> 
> So, I was shipping a patch in the Debian packages for pykolab, that would
> cause setup-kolab to use 'setup-ds' rather than 'setup-ds-admin', but Paul
> decided to drop it (#1040[1], [2])
> 
> [1] https://bugzilla.kolabsys.com/show_bug.cgi?id=1040
> [2]
> http://git.kolabsys.com/apt/pykolab/commit/?id=ed1370beb1024261fdcb2d6e97c9
> 7e186daac863
> 
> > The central parts are:
> > > Could not import LDIF file '/tmp/ldifUN9JLt.ldif'.  Error: 65280.  
Output:
> > importing data ...
> > 
> > > [25/Sep/2012:11:50:56 +0000] - mkdir_p /var/lib/dirsrv/slapd-kolab:
> > > error
> > 
> > -5966 (Access Denied.)
> > 
> > > [25/Sep/2012:11:50:56 +0000] - Can't start because the database
> > > directory
> > 
> > "/var/lib/dirsrv/slapd-kolab/db" either doesn't exist, or is not
> > accessible
> > 
> > > [25/Sep/2012:11:50:56 +0000] - ERROR: Failed to init database (error -1:
> > Unknown error: -1)
> > 
> > > Error: Could not create directory server instance 'kolab'.
> > > Exiting . . .
> > > Log file is '/tmp/setupIdMLnz.log'
> > 
> > '/tmp/ldifUN9JLt.ldif' does not exist and 'ls -la /var/lib/dirsrv/' gives
> > 
> > >drwxrwx---  5 nobody nobody 4096 Sep 25 11:50 slapd-kolab
> 
> I reckon the default user/group account for the directory server was
> patched (in the debian packaging) to be 'dirsrv', not 'nobody'.
Correct. Solved for now by 'chown -R nobody:nobody /var/lib/dirsrv/', but 
maybe it is more sustainable to use user/group "dirsrv" during setup-kolab?

Would that interfere anywhere else?

Sincerely, Joh

_______________________________________________
Kolab-devel mailing list
[email protected]
https://www.intevation.de/mailman/listinfo/kolab-devel

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
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=3Pol
-----END PGP SIGNATURE-----