From: | Michail Nikolaev <michail(dot)nikolaev(at)gmail(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: CPU time for pg_stat_statement |
Date: | 2022-05-20 21:32:47 |
Message-ID: | CANtu0og2w9WSknTH17VnRJ5jDyv9bzSHGCmjRxnBDWcEDnNvKA@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hello, Tom.
> This is a pretty broad claim to make on the basis of one undocumented
> test case on one unmentioned platform.
I'll try to use pg_stat_kcache to check the difference between Wall
and CPU for my case.
> On what grounds do you claim getrusage will be better? One thing we
> can be pretty certain of is that it will be slower, since it has to
> return many more pieces of information. And the API for it only allows
> time info to be specified to microseconds, versus nanoseconds for
> clock_gettime, so it's also going to be taking a precision hit.
My idea was to not replace wall-clock (clock_gettime) by cpu-clock (getrusage).
I think about adding getrusage as an additional column (with flag to
enable actual measuring).
Looks like I need to be more precise in words :)
It is just two different clocks - and sometimes you need physical
time, sometimes CPU time (and sometimes, for example, amount of WAL
written).
Best regards,
Michail.
From | Date | Subject | |
---|---|---|---|
Next Message | Nathan Bossart | 2022-05-20 21:45:19 | Re: Add --{no-,}bypassrls flags to createuser |
Previous Message | Michail Nikolaev | 2022-05-20 21:21:49 | Re: CPU time for pg_stat_statement |