Hello Everyone,
I would like to kindly ask for your help regarding a strange situation I meet yesterday. It might not be a postgres
issuebut I run out of ideas on what to test or check and so far I could not reproduce the behavior.
I have 2 (master - slave) postgres 9.0.3 databases running on SLES 11 PATCHLEVEL 1 (x86_64). For the slave I am using a
statementbased replication (terracotta)...
I have a cron that runs at 1:00am and checks if the dbs are in sync (it exports the data on a text file and makes a
diffon the files generated on the 2 dbs).
2 days ago I've got differences for a table and I started to investigate the problem.
Postgres log is enabled on both dbs and I could compare the statements. The same statements appear in postgres log on
bothdbs in the same order, no errors or weird stuff. So the dbs should be in sync.
That table holds temporary information for an item. So there are inserts and updates on that table for an item (key1)
andwhenever that item expires all the data for that item gets deleted:
insert into mytable (key1, key2, keyval) values (1019, 1, 50);
insert into mytable (key1, key2, keyval) values (1019, 2, 100);
insert into mytable (key1, key2, keyval) values (1019, 3, 10);
update mytable set keyval = 51 where key1 =1019 and key2 = 1;
update mytable set keyval = 102 where key1 =1019 and key2 = 2;
insert into mytable (key1, key2, keyval) values (1019, 4, 10);
...
delete from mytable where key1 = 1019;
