On Wed, 12 Feb 2020 at 12:36, imai.yoshikazu@fujitsu.com
<imai.yoshikazu@fujitsu.com> wrote:
> It seems performance difference is big in case of read only tests. The reason is that write time is relatively longer
thanthe
> processing time of the logic I added in the patch.
That's going to be a pretty difficult performance hit to justify.
Can we buffer collected wait events locally and spit the buffer to the
stats collector at convenient moments? We can use a limited buffer
size with an overflow flag, so we degrade the results rather than
falling over or forcing excessive stats reporting at inappropriate
times.
I suggest that this is also a good opportunity to add some more
tracepoints to PostgreSQL. The wait events facilities are not very
traceable right now. Exposing some better TRACE_POSTGRESQL_
tracepoints (SDTs) via probes.d would help users collect better
information using external tools like perf, bpftrace and systemtap.
That way we have a zero-overhead-when-unused option that can also be
used to aggregate the information per-query, per-user, etc.
(I really need to add a bunch more tracepoints to make this easier...)
--
Craig Ringer http://www.2ndQuadrant.com/
2ndQuadrant - PostgreSQL Solutions for the Enterprise
