Hi, On Tue, 31 Oct 2023 at 16:57, Nazir Bilal Yavuz <byavu...@gmail.com> wrote: > On Thu, 26 Oct 2023 at 09:28, Michael Paquier <mich...@paquier.xyz> wrote: > > > > And perhaps just putting that everything that calls > > pgstat_count_io_op_time() under track_io_timing is just natural? > > What's the performance regression you would expect if both WAL and > > block I/O are controlled by that, still one would expect only one of > > them? > > I will check these and I hope I will come back with something meaningful.
I applied the patches on upstream postgres and then run pgbench for each available clock sources couple of times: # Set fsync = off and track_io_timing = on # pgbench -i -s 100 test # pgbench -M prepared -c16 -j8 -f <( echo "SELECT pg_logical_emit_message(true, \:client_id::text, '1234567890');") -T60 test Results are: ╔═════════╦═══════════════════════════════╦════════╗ ║ ║ track_wal_io_timing ║ ║ ╠═════════╬═══════════════╦═══════════════╬════════╣ ║ clock ║ on ║ off ║ change ║ ║ sources ║ ║ ║ ║ ╠═════════╬═══════════════╬═══════════════╬════════╣ ║ tsc ║ ║ ║ ║ ║ ║ 514814.459170 ║ 519826.284139 ║ %1 ║ ╠═════════╬═══════════════╬═══════════════╬════════╣ ║ hpet ║ ║ ║ ║ ║ ║ 132116.272121 ║ 141820.548447 ║ %7 ║ ╠═════════╬═══════════════╬═══════════════╬════════╣ ║ acpi_pm ║ ║ ║ ║ ║ ║ 394793.092255 ║ 403723.874719 ║ %2 ║ ╚═════════╩═══════════════╩═══════════════╩════════╝ Regards, Nazir Bilal Yavuz Microsoft
