Hi, I'm curious why the maximum size of the sync_group array in remote_worker.c is only 100 events. It seems odd and unpractical that slon doesn't obey the "-g" option. I had to recompile slon today with an increased size in order to get slony to "catch up" after event processing was halted due to my "post-merge re-replication" problem.
I increased the size of the array and the corresponding if statement conditions to 2000. While this is obviously a very large amount of data, the slon daemons didn't inflate due to the incremental fetches from the cursor. Instead, this reduced the number of index/table scans over sl_log_1 by 20x with only a 2x time penalty. Slony changed from falling behind at a dramatic rate to catching up on 3 million transactions in only a few hours. Regards, Michael _______________________________________________ Slony1-general mailing list [email protected] http://gborg.postgresql.org/mailman/listinfo/slony1-general
