Thu Nov 22 04:52:05 PST 2007
- Previous message: [Slony1-general] Patch to have a defined table lock order
- Next message: [Slony1-general] Re: PB with Failover and switchover
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hello, I am using Slony 1.2.12 with Postgresql 8.2.5 I have 2 nodes replicating a small database. The replication works fine but when I try to use failover and switchover I got an error. Failover and switchover were working perfectly before but now I can't use these features anymore, even I downgrade to Slony 1.2.9. Can you please help to find the source of this error? Thanks in advance For the Failover, here is my script : ######################### cluster name =3D my_cluster; node 1 admin conninfo=3D'host=3Dnode1 dbname=3Dbdd user=3Dslony port=3D543= 2'; node 2 admin conninfo=3D'host=3Dnode2 dbname=3Dbdd user=3Dslony port=3D543= 2'; try { failover (id =3D 1, backup node =3D 2); } on error { echo 'Failure to fail node 1 over to 2'; exit 1; } echo 'Replication sets originating on 1 failed over to 2'; try { drop node (id =3D 1, event node =3D 2); } on error { echo 'Failed to drop node 1 from cluster'; exit 1; } echo 'dropped node 2 cluster'; ######################### And here is the error message that appears in my console ( there is nothing in log file): ######################### <stdin>:6: PGRES_FATAL_ERROR select "_my_cluster".failedNode(1, 2); - ERROR: relation "PartInd_my_cluster_sl_log_2-" already exists CONTEXT: SQL statement "create index "PartInd_my_cluster_sl_log_2-node-2" on "_my_cluster".sl_log_2 USING btree(log_xid "_my_cluster".xxid_ops) where (log_origin =3D 2);" PL/pgSQL function "addpartiallogindices" line 34 at execute statement SQL statement "SELECT "_my_cluster".addPartialLogIndices()" PL/pgSQL function "failednode" line 158 at perform <stdin>:8: Failure to fail node 1 over to 2 ######################### When I try to do a switchover using this script : ######################### cluster name =3D my_cluster; node 1 admin conninfo=3D'host=3Dnode1 dbname=3Dbdduser=3Dslony port=3D5432'; node 2 admin conninfo=3D'host=3Dnode2 dbname=3Dbdd user=3Dslony port=3D5432= '; echo 'Moving set 1 to node 2'; lock set (id=3D1, origin=3D1); echo 'Set locked'; wait for event (origin =3D 1, confirmed =3D 2); echo 'Moving set'; move set (id=3D1, old origin=3D1, new origin=3D2); echo 'Set moved - waiting for event to be confirmed by node 2'; wait for event (origin =3D 1, confirmed =3D 2); echo 'Confirmed'; ######################### The slonik script output is locked on "<stdin>:11: Set moved - waiting for event to be confirmed by node 2" and I got an error message equivalent to the one I got for failover but in log file. I also ######################### "select "_hwi06intrsifoisusgbd1_2_hwi06intrsifoisusgbd2".cleanupNodelock(); insert into "_hwi06intrsifoisusgbd1_2_hwi06intrsifoisusgbd2".sl_nodelock values ( 2, 0, "pg_catalog".pg_backend_pid()); " - ERROR: duplicate key violates unique constraint "sl_nodelock-pkey" ######################### On node1, the sl_nodelock table contains : 1 0 21582 2 1 21682 2 2 21684 2 3 21865 On node2, the sl_nodelock table contains : 1 1 25103 2 0 25164 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://lists.slony.info/pipermail/slony1-general/attachments/20071122/= fac84a21/attachment-0001.html
- Previous message: [Slony1-general] Patch to have a defined table lock order
- Next message: [Slony1-general] Re: PB with Failover and switchover
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list