-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stéphane A. Schildknecht a écrit :
> Hi,
> 
> I'm trying to upgrade from slony 1.2 branch to 2.0 branch.
> 
> configure, make and make install went fine.
> 
> I then tried to execute the update functions part, but got an error message.
> 
> Here's what I did :
> 
> opt/appli/slony1-2.0/bin/slonik <<_EOF_
>         #--
>         # Inclusion du fichier <preamble>
>         #--
>         include </opt/appli/slony1/etc/test2.preamble>
> 
>         #--
>         # Mise à jours des schémas
> 
>         #--
>         try{
>            update functions ( id=1 );
>            update functions ( id=2 );
>            update functions ( id=3 );
>         }
>         on error { echo 'Mise à jour schémas Slony : KO' ;}
>         on success { echo 'Mise à jour schémas Slony : OK' ; }
> 
> _EOF_
> 
> And the message I get :
> <stdin>:10: loading of file /usr/share/postgresql/8.3/slony1_funcs.sql:
> PGRES_FATAL_ERROR ERREUR:  le type « _test2.vactables » n'existe pas
> ERREUR:  le type « _test2.vactables » n'existe pas
> <stdin>:15: Mise à jour schémas Slony : KO
> 
> PostgreSQL is version 8.3.5, slony version was 1.2.14.
> 
> In fact, I did some search, and I guess trouble is at loading slony1_func.sql,
> while trying to create fonction TablesToVacuum () as it returns a setof
> @NAMESPACE at .vactables, but, that type can't exist, as it is created by a call
> to upgradeSchema.
> 
> Should the loading of that file being done in two times ? Once for creation of
> the type, second for creation of functions using it ?
> 
> Best regards,

OK, I did try to do it in 2 phases. First I commented out the creation of
function TableToVacuum. The upgrade was quite successful. But, when restarting
slon daemons, I got these repeting errors :

2008-12-17 15:07:46 CET ERREUR:  la colonne « ev_snapshot » n'existe pas au
caractère 50
2008-12-17 15:07:46 CET INSTRUCTION :  select ev_origin, ev_seqno,
ev_timestamp,        ev_snapshot,
"pg_catalog".txid_snapshot_xmin(ev_snapshot),
"pg_catalog".txid_snapshot_xmax(ev_snapshot),        ev_type,        ev_data1,
ev_data2,        ev_data3, ev_data4,        ev_data5, ev_data6,
ev_data7, ev_data8 from "_test2".sl_event e where (e.ev_origin = '3' and
e.ev_seqno > '325') order by e.ev_origin, e.ev_seqno limit 40

In fact, definition of sl_event has changed, and it seems like my existing
_test2 schemas have had no information of that change.

Is it something I should have discovered if I did read some piece of
documentation I'm not  aware of ?
Documentation of slony upgrade does not tell anything about a change in
procedure since I did last upgrade my slony installation.


Best regards,
- --
Stéphane Schildknecht
PostgreSQLFr - http://www.postgresql.fr
Dalibo - http://www.dalibo.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFJSQs2A+REPKWGI0ERAssQAJ4/z4Q9ExQ+nrrZ0BDJXh0rSBS+KgCeKRqX
03xy8zvvKUiFBB31gKh2UQo=
=K9nr
-----END PGP SIGNATURE-----