On Thu, 2012-05-24 at 07:29 -0400, Stephen Gallagher wrote:
On Thu, 2012-05-24 at 08:19 +0200, Jan Zelený wrote:
> > On Wed, 2012-05-23 at 12:03 +0200, Jan Zelený wrote:
> > > > On Thu, May 10, 2012 at 10:57:23PM +0200, Jan Zeleny wrote:
> > The sysdb upgrade script will segfault if any users in the database are
> > lacking memberOf links. This can happen if a user was requested via
> > getpwnam() or getpwuid() without an initgroups() call.
> >
> > The attach patch should be squashed into patch 122.
>
> Nice catch, thank you
>
> > Additionally, I was trying to do some real-world comparisons but I
> > wasn't seeing any improvement (this was against an RFC 2307 server, not
> > RFC2307bis or FreeIPA, so that may be relevant).
> >
> > Can you verify whether we expect to see improvement on such systems, or
> > only for RFC2307bis?
>
> I don't know why we shouldn't expect any improvements for RFC2307. Can you
> post me the scenario you tested?
Replied privately with confidential data.
Ok, we sorted out that the lack of performance improvement was due to
high latency to the LDAP server and not actually the ghost changes. When
I moved to a local LDAP server, performance has actually improved by
approximately 300%.
So, with squashing in my patch for the upgrades, I call this an ack.
Attaching all the patches (with the upgrade patch squashed in) to make
pushing easier.