Hi,
I’m using PostgreSQL 10 on a mac pro running macOS Mojave. My version info for PGAdmin4 is provided below.
I installed EDB standard PostgreSQL version 10 on the mac pro as I have done many times before. I’m going to be using this implementation to perform ETL, List Management, GIS operations using Tiger and OSM, so I’m isolating the operations for each of those processes using Tablespaces. And I’m performing the configuration of the tablespaces using the PGAdmin4 web tool in chrome.
I added a larger group of tablespaces using the PGAdmin4 tool over a TCP connection to a Ubuntu server on my network with no problems similar to this. In that case, only the connected PostgreSQL configuration was affected. The database on the connected Ubuntu system is a separate installation of PostgreSQL, whose role is to support the Pentaho BA Server and its PostgreSQL DB repository. And the new tablespaces created in that connected system’s LVM disks seem to act in a reasonable manner and do not display this duplication.
However, in addition to the connected Server on the Ubuntu machine, I have three local server instances showing in the PGAdmin interface on the mac pro. These are the main local PostgreSQL server instance, and two other local server instances (ETL and GIS) that isolate development from the Ubuntu system and the local Server. All three of these local server instances are looking at the same PostgreSQL installation and PGData folder on the Mac Pro.
When I add tablespaces to any of these three local server instances, each of the local server instances gets a copy of each of the new tablespaces. These become visible in the PGAdmin4 Tablespace folder of the server instance. This outcome is puzzling because the PostgreSQL 10 docs state that Tablespaces must be unique across the Database Cluster.
I have found no information indicating that Postgres might elect scope the Tablespace names by Server instance. But I’m not even sure whether the ‘Server’ concept is a PostgreSQL concept or a PGAdmin4 concept. So, at this point, I can’t determine whether this is a PostgreSQL misunderstanding or a PGAdmin4 issue. My reading of PostgreSQL docs indicates that a Cluster encompasses Databases, not Servers, but maybe I haven’t run across the distinguishing facts on that.
Nonetheless, this duplication is alarming. Does any of this sound familiar?
Thanks,
Kimball Johnson
Product Manager at Engagile.io
Version
3.1
Copyright
Copyright 2013 - 2018, The pgAdmin Development Team
Python Version
3.6.5 (default, Jun 15 2018, 09:47:54) [GCC 4.2.1 Compatible Apple LLVM 7.0.2 (clang-700.1.81)]
Flask Version
0.12.2
Application Mode
Desktop
Current User
pgadmin4@pgadmin.org