On 2018-08-13 11:50:41 -0300, Alvaro Herrera wrote:
> On 2018-Aug-11, Andres Freund wrote:
>
> > On 2018-08-11 00:46:25 -0700, Andres Freund wrote:
> > > Below you can find the bt full showing a bunch of nested invalidations. Looking.
> >
> > Hm. I wonder if the attached fixes the issue for you. If it's this I
> > don't understand why this doesn't occur on older branches...
> >
> > I've not yet been able to reliably reproduce the issue without the
> > patch, so I'm not sure it means much that it didn't reoccur afterwards.
>
> Well, the line you're patching was introduced in 2ce64caaf793 (looking
> at pg10), and was not replacing a line with the exact same mistake.
I'm not clear on what you're trying to say? My problem is that Thomas'
backtraces don't show involvement by that function, which makes it less
likely that it matters, no?
Greetings,
Andres Freund