Gregory Stark wrote:
> "Gregory Stark" <stark@enterprisedb.com> writes:
>
>> On the other hand we can't just ignore all vacuums because someone could issue
>> a manual vacuum inside a transaction (I think?).
>
> Doh, ignore this. sigh.
I started from scratch to put up a test case. I cannot trigger "ERROR:
item pointer (0,1) already exists" again as the deadlock issue reported
by Gregory Stark hit me every time. Is this fixed in RC1? Can I retry
with that?
Only good news is that I think I found the CLUSTER issue:
It was no GIST index I created, I accidentally created a BTREE index:
http://theendofthetunnel.de/cluster.txt
If it does help anything, the data and setup:
8.3-b4 build from source
./configure --prefix=/usr/local --enable-thread-safety --with-perl
--with-openssl --with-libxml --with-libxslt
initdb line:
initdb -D /data/pgsql --locale='de_DE.utf8' --lc-collate='C'
Only listen_address and pg_hba.conf was touched.
Please get the -Fc dump (37MB) from:
http://theendofthetunnel.de/dump.bin
http://theendofthetunnel.de/glob.sql
--
Best regards,
Hannes Dorbath