Thank you for your answer, but I don't have access to this server any more and have to just interpret and pick some parameters from test results, but your advice about tar trick is something I'm gonna try for sure in next test. So, because I don't have any chance to do some more tests, should I change checkpoint_segments parameter? My colleagues prefer old setting as shown below, because of maintenance reasons, but I still would like to convince them to much higher setting. 30 segments for machine like that seems to be too humble. checkpoint_segments = 30 checkpoint_timeout = 8min The rest of config looks like this: shared_buffers=2GB temp_buffers=128MB max_files_per_process=1000 work_mem=384MB maintenance_work_mem=10240MB effective_io_concurrency=1 synchronous_commit=on wal_buffers=16MB wal_writer_delay=200ms commit_delay=0 commit_siblings=5 random_page_cost=1.0 cpu_tuple_cost= 0.01 effective_cache_size=450GB geqo_threshold=12 geqo_effort=5 geqo_selection_bias=2.0 join_collapse_limit=8 Any ideas about rest of config? Maybe those connected with write operations? -- View this message in context: http://postgresql.1045698.n5.nabble.com/Optimal-checkpoint-setting-tp5822359p5822951.html Sent from the PostgreSQL - general mailing list archive at Nabble.com. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
