Thread: [HACKERS] PATCH: pgbench - break out timing data for initialization phases
This patch is based against 'master'.
It adds collection of timing data for the initialization option phases and
reporting at the end of initialization.
Before patch:
.
.
.
199900000 of 200000000 tuples (99%) done (elapsed 158.96 s, remaining 0.08 s)
200000000 of 200000000 tuples (100%) done (elapsed 159.04 s, remaining 0.00 s)
vacuum...
set primary keys...
done.
After patch:
.
.
.
199900000 of 200000000 tuples (99%) done (elapsed 161.48 s, remaining 0.08 s)
200000000 of 200000000 tuples (100%) done (elapsed 161.56 s, remaining 0.00 s)
vacuum...
set primary keys...
total time: 316.03 s (insert 161.60 s, commit 0.64 s, vacuum 60.77 s, index 93.01 s)
done.
doug
--
Doug Rady
Amazon Aurora PostgreSQL
radydoug@amazon.com
Attachment
Re: [HACKERS] PATCH: pgbench - break out timing data for initializationphases
From
Fabien COELHO
Date:
Hello Doug, > total time: 316.03 s (insert 161.60 s, commit 0.64 s, vacuum 60.77 s, index 93.01 s) Definitely interesting. There is a "ready for committers" patch in the CF which extensively rework the initialization: it becomes customizable, and this approach may not work as is after that... Maybe you could investigate how it may be implemented on top of that? Either show times when the phases are performed computed, or maybe use some auxiliary data structure to keep the information (hmmm...). -- Fabien. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] PATCH: pgbench - break out timing data forinitialization phases
From
Michael Paquier
Date:
On Tue, Sep 26, 2017 at 3:07 AM, Fabien COELHO <coelho@cri.ensmp.fr> wrote: > > Hello Doug, > >> total time: 316.03 s (insert 161.60 s, commit 0.64 s, vacuum 60.77 s, >> index 93.01 s) > > > Definitely interesting. > > There is a "ready for committers" patch in the CF which extensively rework > the initialization: it becomes customizable, and this approach may not work > as is after that... > > Maybe you could investigate how it may be implemented on top of that? > > Either show times when the phases are performed computed, or maybe use some > auxiliary data structure to keep the information (hmmm...). The patch needs as well a rebase, I am marking it as returned with feedback per lack of activity. When sending a new version, please make sure to: - reply to the feedback which has been provided previously. - register a new version of the patch in the commit fest app. Thanks, -- Michael