On 08/30/2016 10:06 PM, Lukas Slebodnik wrote:
> On (30/08/16 21:22), Jakub Hrozek wrote:
> > On Tue, Aug 30, 2016 at 06:51:43PM +0200, Lukas Slebodnik wrote:
> > > On (30/08/16 13:01), Pavel Březina wrote:
> > > > On 08/19/2016 02:03 PM, Jakub Hrozek wrote:
> > > > > Hi,
> > > > >
> > > > > I'm sending this patch on behalf of Nalin, who attached it
to
> > > > >
https://bugzilla.redhat.com/show_bug.cgi?id=1063278#c9
> > > >
> > > > Ack.
> > > >
> > > > Solves:
> > > >
https://fedorahosted.org/sssd/ticket/1498
> > > Denis wrote a comment into BZ
> > >
> > > in quick testing it works. fall back is very slow, but it does appear
> > > to do so.
> > >
> > > Current version iterate over all addressed and we can contine with next
one
> > > only if the 1st failed.
> > >
> > > Could we try to connect to all addressed at once and use the fastest
> > > (pool should use the fastest connection)
> >
> > Feel free to write a patch. I think Nalin's patch is already an
> > improvement over total breakage we have now.
> The question is what is better.
>
> The broken version or *very slow* version.
> IMHO broken version.
>
> LS
Speed depends on error we get. I tested it on in our vm lab by connecting to
fedorapeople.org and it connected immediately. Without the patch it tries to
connect through ipv6 which doesn't work and it fails completely.
We should accept this patch and create and defer a ticket for your
suggestion.