Stefan Kaltenbrunner <stefan@kaltenbrunner.cc> writes:
> the machine had some issues a week ago or so - but it looks like the
> problem occured first here:
Hmm, what kind of issues, and are you sure they are fixed?
The stack trace looks to me like it is trying to apply the PGTZ setting
that's coming in from the client during startup. Now I could believe a
platform-specific breakage there from Magnus' recent hacking, but the
problem is that *every* backend launched during the regression tests
is going to be doing this, and doing it in the same way with the same
value. It's pretty tough to see why some backends would fail at this
spot and some not ... unless what it is is an intermittent hardware
problem. Is there a version of memtest86 that works on your machine?
regards, tom lane