Hi,
just to give an update on this:
Frederik Ramm wrote:
> Every other night, the process aborts with some strange error message,
> and never at the same position:
[...]
Turns out it *was* a RAM defect on one of the machines. memtest86 ran
for a day and didn't detect it, but when I started making copies of a
100 GB file and m5dsumming it afterwards, I'd get one or two bad copies
per day. I then juggled RAM chips until I had identified the defective
one. As you can imagine, this took a while.
Lesson learnt: memtest86 running for a day and not finding an error does
not mean your RAM is ok!
I still have occasional problems on the *other* machine (if you remember
my original post, I had similar errors on two different machines) and
even massive data copying did not show any error there but at least a
PostGIS bug seems extremely unlikely now (and that's still running stock
packages, not patched).
Bye
Frederik
--
Frederik Ramm www.geofabrik.de
Geofabrik GmbH Handelsregister: HRB Mannheim 703657
Rueppurrer Strasse 4 Geschaeftsfuehrung: Frederik Ramm
76137 Karlsruhe Tel: 0721-1803560-0
frederik.ramm@geofabrik.de Fax: 0721-1803560-9
