Re: Monitoring time of fsyncing WALs

On Mon, Jul 02, 2018 at 11:36:06AM +0800, Craig Ringer wrote:
> On 1 July 2018 at 11:29, Michael Paquier <michael@paquier.xyz> wrote:
>> So at the end, I would like to use the proposed patch and call it a
>> day.  Thoughts?
>>
> Patch looks good.

Thanks Craig for the review!  I have just pushed the previous patch with
the adjustments mentioned.  Regarding enforcing pg_fsync with a wait
event, I think I'll step back on this one.  There are also arguments for
allowing code paths to not have wait events as that's quite low-level,
and for extensions the choice is limited as long as there is no way to
register custom names, if that ever happens, which I am not sure of.

> I'll hijack the thread to add a few more perf/dtrace tracepoints in the WAL
> code, as they were also missing. Proposed rider patch attached.

:)

As that is a separate discussion and as the commit fest has already
begun with many items in the queue, could you begin a new thread and add
this stuff within a new CF entry?  I would say that having more trace
points in this area could be helpful, but let's discuss that
appropriately so as your proposal catches proper attention.

> BTW, we might want to instrument the pgstat_ counter calls
> and pgstat_report_wait_start / pgstat_report_wait_end, but it's easy enough
> to use dynamic tracepoints for those so I haven't yet. Maybe even just
> document them as points of interest.

Perhaps.  Feel free to propose anything you have in mind of course.
--
Michael