Horrible CREATE DATABASE Performance in High Sierra

From: Brent Dearth <brent(dot)dearth(at)gmail(dot)com>
To: pgsql-hackers(at)postgresql(dot)org
Subject: Horrible CREATE DATABASE Performance in High Sierra
Date: 2017-10-02 15:53:12
Message-ID: CADkxhTNv-j2jw2g8H57deMeAbfRgYBoLmVuXkC=YCFBXRuCOww@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

I just recently "upgraded" to High Sierra and experiencing horrendous CREATE
DATABASE performance. Creating a database from a 3G template DB used to
take ~1m but post-upgrade is taking ~22m at a sustained write of around
4MB/s. Occasionally, attempting to create an empty database hangs
indefinitely as well. When this happens, restarting the Postgres server
allows empty database initialization in ~1s.

I had been running on an encrypted APFS volume (FileVault), but after
dropping encryption, saw the tasks drop to about *~15m* per run. Still a
far cry from the previous *~1m* threshold.

A multi-threaded pg_restore seems to sustain writes of ~38M/s and completes
in about the same time as pre-upgrade (Sierra), so I'm not sure it's
entirely related to APFS / disk IO.

I've completely rebuilt the Postgres data directory, re-installed Postgres
(Postgres.app 2.0.5) etc. I don't have any reasonable explanation for what
could have broken so catastrophically.

Coworker has seen the exact same issue. Has anyone else experienced this
yet or have any insight as to what could be happening?

Thanks in advance!

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Alvaro Herrera 2017-10-02 15:54:08 Re: [PATCH] Assert that the correct locks are held when calling PageGetLSN()
Previous Message Konstantin Knizhnik 2017-10-02 15:49:48 Issues with logical replication