Andres Freund wrote on 2018-03-02: > Yea, I misread the diff to think you added a conflicting version. Due > to: > -DATA =3D pg_stat_statements--1.4.sql pg_stat_statements--1.4--1.5.sql \ > +DATA =3D pg_stat_statements--1.5.sql pg_stat_statements--1.4--1.5.sql \
> and I'd checked that 1.5 already exists. But you just renamed the file, > presumably because it's essentially rewriting the whole file? I'm not > sure I'm a big fan of doing so, because that makes testing the upgrade > path more work. You're right about 1.5 already existing. I wasn't sure about the versioning policy for extensions and seeing as version 1.5 only changed a few grants I quasi reused 1.5 for my intentions. > What workload did you run? read/write or readonly? This seems like a > feature were readonly makes a lot more sense. But ~1800 tps strongly > suggests that's not what you did? I'm sorry I forgot to mention this; I ran all tests as read-write. > > With pg_stat_statements on, the latter test (10 minutes) resulted in 1833 > > tps, while the patched version resulted in 1700 tps, so a little over 7% > > overhead? Well, the "control run", without pg_stat_statements delivered > > only 1806 tps, so variance seems to be quite high. > That's quite some overhead, I'd say. Yes, but I wouldn't give a warranty that it is neither more nor less overhead than 7%, seeing as for my testing, the tps were higher for (unmodified) pgss enabled vs no pgss at all. > > If anybody has any recommendations for a setup that generates less > > variance, I'll try this again. > I'd suggest disabling turboost, in my experience that makes tests > painful to repeat, because it'll strongly depend on the current HW > temperature. This might be a problem for average systems but I'm fairly certain this isn't the issue here. I might try some more benchmarks and will in particular look into running read-only tests, as the aforementioned 840 EVO SSD ist -comparatively speaking- pretty slow. Do you have any recommendations as to what constitutes adequate testing times regarding pgbench? Kind regards Julian
