Re: Additional LWLOCK_STATS statistics

From: Jesper Pedersen <jesper(dot)pedersen(at)redhat(dot)com>
To: Robert Haas <robertmhaas(at)gmail(dot)com>
Cc: "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Additional LWLOCK_STATS statistics
Date: 2015-09-15 19:30:02
Message-ID: 55F871BA.4060404@redhat.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 09/15/2015 03:11 PM, Robert Haas wrote:
>> If there is an interest I'll add the patch to the next CommitFest.
>>
>> Thanks for considering, and any feedback is most welcomed.
>
> Seems neat, but I can't understand how to read the flame graphs.
>

X-axis is sort of "up in the air" with flame graphs -- similar call
stacks are grouped together, and here it is the queue size.

Y-axis is the lock queue size -- e.g. CLogControlLock is "max'ed" out,
since there is a queue size of 80 with the -c/-j 80 run.

The width of each column shows how the locks are "distributed" within
the report. Mouse over on the lock name gives the % of the entire
report, f.ex. CLogControlLock has 74% of spin delays. Note, that some of
the reports are filtered in order to eliminate "unnecessary" information
-- see README.

You would need to zoom into some of the information - left click - and
the reset the zoom afterwards.

I can change the reports if something is more helpful.

Flame graphs for CPU profiling are more common, like

https://dl.dropboxusercontent.com/u/99317937/postgresql/96S-20150808-r30t-17353.svg

Thanks for looking at the patch.

Best regards,
Jesper

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Jeff Janes 2015-09-15 19:30:40 Re: pgbench progress with timestamp
Previous Message Adam Brightwell 2015-09-15 19:18:21 Re: row_security GUC, BYPASSRLS