Vivek Khera <khera(at)kcilink(dot)com> writes:
> Restore of a significanly big database (~19.8GB restored) shows nearly
> no time difference depending on sort_mem when checkpoint_segments is
> large. There are quite a number of tables and indexes. The restore
> was done from a pg_dump -Fc dump of one database.
I was just bugging Marc for some useful data, so I'll ask you too:
could you provide a trace of the pg_restore execution? log_statement
plus log_duration output would do it. I am curious to understand
exactly which steps in the restore are significant time sinks.
> I notice during the restore that the disk throughput triples during
> the checkpoint.
Hm, better make sure the log includes some indication of when
checkpoints happen.
regards, tom lane