we have a replcation environment. This composed for 3 snapshot and 1 master. A modification in a snapshot site is pushed to the master site, and this modification is replied in the another snapshot.
But now, an insert in a table (member) of the snapshot group (there are 4 members), the new data are pushed to the master, but this modifications are delete (lose) in the origin table and are not replicated in the other snapshot.
The transactions are in fact being pushed back to the master site and the transactions are successfully purged from deftran (DEF$_AQCALL), but the insert in the snapshot is lost, and the changes are not being reflected in the others snapshot tables.
I have read the Metalink note 1040975.6, and the parameter PUSH_DEFERRED_RPC is TRUE. In the SYSTEM.DEF$_AQERROR table in the master, there is a lot of errors (around 3.000.000 rows) and in the MLOG$_... table in the master, the inserts in the snapshot aren't reflected.
In Alert log I don´t see nothing.
The refresh job in the snapshot is:
If I do a manual complete refresh in the snapshot of the table that is not refresh by the job, this complete refresh is ejecuted successfully.
Could be a solution to delete the table of the replication group in the master and then incorporate this again in the group and then to execute a complete refresh of this table in the snapashot to synchronize?
Do you have conflict resolution setup at the master site ... Could it
be that the master is rejecting the "push" because of foreign keys
that are not set up as "deferable"??? If these are the cases, the
transaction will be rolled back out of the snapshots ...
Something on the Master site is rejecting the trx...