Resetting PostgreSQL
My original answer only included the troubleshooting steps below, and a workaround. I now decided to properly fix it via brute force by removing all clusters and reinstalling, since I didn't have any data there to keep. It was something along these lines, on my Ubuntu 21.04 system:
sudo pg_dropcluster --stop 12 main
sudo pg_dropcluster --stop 14 main
sudo apt remove postgresql-14
sudo apt purge postgresql*
sudo apt install postgresql-14
Now I have:
$ pg_lsclusters
Ver Cluster Port Status Owner Data directory Log file
14 main 5432 online postgres /var/lib/postgresql/14/main /var/log/postgresql/postgresql-14-main.log
And sudo -u postgres psql works fine. The service was started automatically but it can be done manually with sudo systemctl start postgresql.
Incidentally, I can recommend the PostgreSQL docker image, which eliminates the need to bother with a local installation.
Troubleshooting
Although I cannot provide an answer to your specific problem, I thought I'd share my troubleshooting steps, hoping that it might be of some help. It seems that you are on Mac, whereas I am running Ubuntu 21.04, so expect things to be different.
This is a client connection problem, as noted by section 19.3.2 in the docs.
The directory in my error message is different:
$ sudo su postgres -c "psql"
psql: error: connection to server on socket "/var/run/postgresql/.s.PGSQL.5432" failed: No such file or directory
Is the server running locally and accepting connections on that socket?
I checked what unix sockets I had in that directory:
$ ls -lah /var/run/postgresql/
total 8.0K
drwxrwsr-x 4 postgres postgres 160 Oct 29 16:40 .
drwxr-xr-x 36 root root 1.1K Oct 29 14:08 ..
drwxr-s--- 2 postgres postgres 40 Oct 29 14:33 12-main.pg_stat_tmp
drwxr-s--- 2 postgres postgres 120 Oct 29 16:59 14-main.pg_stat_tmp
-rw-r--r-- 1 postgres postgres 6 Oct 29 16:36 14-main.pid
srwxrwxrwx 1 postgres postgres 0 Oct 29 16:36 .s.PGSQL.5433
-rw------- 1 postgres postgres 70 Oct 29 16:36 .s.PGSQL.5433.lock
Makes sense, there is a socket for 5433 not 5432. I confirmed this by running:
$ pg_lsclusters
Ver Cluster Port Status Owner Data directory Log file
12 main 5432 down,binaries_missing postgres /var/lib/postgresql/12/main /var/log/postgresql/postgresql-12-main.log
14 main 5433 online postgres /var/lib/postgresql/14/main /var/log/postgresql/postgresql-14-main.log
This explains how it got into this mess on my system. The default port is 5432, but after I upgraded from version 12 to 14, the server was setup to listen to 5433, presumably because it considered 5432 as already taken. Two alternatives here, get the server to listen on 5432 which is the client's default, or get the client to use 5433.
Let's try it by changing the client's parameters:
$ sudo su postgres -c "psql --port=5433"
psql (14.0 (Ubuntu 14.0-1.pgdg21.04+1))
Type "help" for help.
postgres=#
It worked! Now, to make it permanent I'm supposed to put this setting on a psqlrc or ~/.psqlrc file. The thin documentation on this (under "Files") was not helpful to me as I was not sure on the syntax and my attempts did not change the client's default, so I moved on.
To change the server I looked for the postgresql.conf mentioned in the documentation but could not find the file. I did however see /var/lib/postgresql/14/main/postgresql.auto.conf so I created it on the same directory with the content:
port = 5432
Restarted the server: sudo systemctl restart postgresql
But the error persisted because, as the logs confirmed, the port did not change:
$ tail /var/log/postgresql/postgresql-14-main.log
...
2021-10-29 16:36:12.195 UTC [25236] LOG: listening on IPv4 address "127.0.0.1", port 5433
2021-10-29 16:36:12.198 UTC [25236] LOG: listening on Unix socket "/var/run/postgresql/.s.PGSQL.5433"
2021-10-29 16:36:12.204 UTC [25237] LOG: database system was shut down at 2021-10-29 16:36:12 UTC
2021-10-29 16:36:12.210 UTC [25236] LOG: database system is ready to accept connections
After other attempts did not succeed, I eventually decided to use a workaround: to redirect the client's requests on 5432 to 5433:
ln -s /var/run/postgresql/.s.PGSQL.5433 /var/run/postgresql/.s.PGSQL.5432
This is what I have now:
$ ls -lah /var/run/postgresql/
total 8.0K
drwxrwsr-x 4 postgres postgres 160 Oct 29 16:40 .
drwxr-xr-x 36 root root 1.1K Oct 29 14:08 ..
drwxr-s--- 2 postgres postgres 40 Oct 29 14:33 12-main.pg_stat_tmp
drwxr-s--- 2 postgres postgres 120 Oct 29 16:59 14-main.pg_stat_tmp
-rw-r--r-- 1 postgres postgres 6 Oct 29 16:36 14-main.pid
lrwxrwxrwx 1 postgres postgres 33 Oct 29 16:40 .s.PGSQL.5432 -> /var/run/postgresql/.s.PGSQL.5433
srwxrwxrwx 1 postgres postgres 0 Oct 29 16:36 .s.PGSQL.5433
-rw------- 1 postgres postgres 70 Oct 29 16:36 .s.PGSQL.5433.lock
This means I can now just run psql without having to explicitly set the port to 5433. Now, this is a hack and I would not recommend it. But in my development system I am happy with it for now, because I don't have more time to spend on this. This is why I shared the steps and the links so that you can find a proper solution for your case.
