We were able to reduce the volume of generated WAL files. In some cases we use UNLOGGED tables to assist in WAL reduction. But I still find it strange the process of recovery in the standby database use little CPU. Today the process uses between 20 and 50% of a CPU, not using all the processing that the machine offers (20 CPUs, RAID 0 on SSD).
Yes that's correct because we have just a single wal-receiver process on standby.
Is there any configuration that allows Postgres to use more processing at the time of the application of WAL files in the StandBy database?
Unfortunately no as I mentioned before...
Did you already run a 'perf top' or 'strace' on recovery process to try understand what happen?
Sorry but I didn't read the entire thread.
Regards,
--
Fabrízio de Royes Mello Timbira - http://www.timbira.com.br/ PostgreSQL: Consultoria, Desenvolvimento, Suporte 24x7 e Treinamento
