Postgres.app Could not start on port 5432

Question

I'm using http://postgresapp.com. In the menubar it gives the error " Could not start on port 5432." Similarly if I try to start the server from the terminal, I get:

psql: could not connect to server: No such file or directory
    Is the server running locally and accepting
    connections on Unix domain socket "/tmp/.s.PGSQL.5432"?

I also ran pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start and got the output server starting but still get the same errors when connecting to psql.

My first guess is that port 5432 is taken by something else on your mac. By default psql will not use the port, but instead use the socket file (`/tmp/.s.PGSQL.5432`) as you mentioned. You can force it to use the port with `psql -h localhost`. Try checking a couple of things: 1 is anything else using port 5432. 2: check that your postgres (OS) user has access to create files in /tmp/. — Philip Couling, Aug 19 '12 at 16:37
I see something in /tmp/ that is .s.PSQL.5433 How can I change that to 5432? — Chris, Aug 19 '12 at 17:01
I'm having the same issue. I think it's related to an update of pro applications from Apple, because that's the only change I made, then I started getting this error. I've tried uninstalling and re-installing postgresql; uninstalling and re-installing the pg gem; and a few other configuration tweaks, but no results. Still working on it, but hope to see a solution soon. I will let you know if I make any progress. — Brian McDonough, Aug 19 '12 at 17:37
In my case, I had dBeaver running connected to my postgresql instance. I made a config change and clicked "Stop" and "Start" and was perplexed to get the "port in use" error. But the "lsof -i :5432" suggested below showed dBeaver had the port tied up. Killed dBeaver and postgresql started up just fine. — Eric Hill, Feb 04 '21 at 20:06

Max von Hippel · Answer 1 · 2019-05-09T00:44:16.843

89

If you are running your server on a macOS machine and installed Postgres through Homebrew, you can stop the current instance like this:

brew services stop postgresql

Then click the Elephant in the native menu-bar at the top of the screen and it should successfully startup.

edited May 09 '19 at 00:44

answered Oct 18 '16 at 05:01

Max von Hippel

2,856
3
29
46

1

where is the menu bar ? :') – Kevin Upton May 09 '19 at 00:17
Good catch! I've clarified which menu bar I meant. Thanks! – Max von Hippel May 09 '19 at 00:44
2

@MaxvonHippel Thank you so much my friend, you really made my day :) – Yavuz Asmalı Nov 26 '20 at 19:11

score 75 · Answer 2 · answered May 06 '17 at 21:43

75

You can stop the process by finding the PID with

lsof -i :5432

and then killing it with

kill -9 <PID>

answered May 06 '17 at 21:43

Некто

1,730
10
17

2

this is the most general method that works for any port. Should be the best answer – mythicalcoder Jul 04 '19 at 07:21
2

When I kill it, again it is running at different PID. please help – Tula Jul 27 '20 at 07:34
@Tula https://unix.stackexchange.com/questions/158194/kill-a-process-that-keeps-restarting – Некто Aug 04 '20 at 19:15
4

I tried this but it would just start running with another PID. What worked for me is `brew services stop postgresql` – Adam Zarn Jul 11 '21 at 19:09

score 34 · Answer 3 · edited Dec 11 '19 at 05:18

34

If you've installed Postgres via another method (for example, from www.postgresql.org) and it's starting automatically at startup, you can prevent that Postgres from starting via the following:

sudo launchctl unload /Library/LaunchDaemons/com.edb.launchd.postgresql-X.X.plist  
sudo rm -f /Library/LaunchDaemons/com.edb.launchd.postgresql-X.X.plist

TIP: use tab to autocomplete after the com.edb.launchd part to figure out what version is loading.

Restart Postgres.app and you should be good to go.

(from http://forums.enterprisedb.com/posts/list/1831.page;jsessionid=70621DC48C99EDE663A6A594B05F1A02#6782)

edited Dec 11 '19 at 05:18

Boris Verkhovskiy

14,854
11
100
103

answered Nov 15 '13 at 20:30

Jamon Holmgren

23,738
6
59
75

thanks a lot. This has worked for me. The issue was that whenever i killed the process listening on 5432, it launched again -_-. Thanks again – Abdellah Stands with Palestine Jul 24 '19 at 13:48
Life saver (need more words to comment this) :D – Kohls Oct 03 '20 at 21:29

score 24 · Answer 4 · edited Jul 14 '15 at 18:26

I was just having this exact issue. When I ran which psql it was pointing at the Postgres client tools installed with Lion:

/usr/bin/psql

Using a hint from Frank Wiles I ran ps auxw | grep post to confirm that postgres was running and that it was running on the right port, that also showed me the postgres.app path:

/Applications/Postgres.app/Contents/MacOS/bin/postgres.

So I edited by .bash_profile to export that directory. On first effort I added it to the end of the path. When I ran echo $PATH I could see that usr/bin was the first thing in the path, and which psql still gave the /usr/bin path. At that point a friend guided me in the right direction:

export PATH="/Applications/Postgres.app/Contents/MacOS/bin:${PATH}"

Start a new terminal window, then run which psql -- it should point to the postgres.app location and psql should fire up the postgres shell. Works fine now.

for 9.3 my path was export PATH="/Applications/Postgres93.app/Contents/MacOS/bin:${PATH}" — LennonR, Jan 15 '14 at 21:07

score 19 · Answer 5 · answered Dec 18 '18 at 01:03

19

This command is a one-liner that instantly kills all PostgresSQL processes.

sudo kill -kill $(sudo lsof -t -i :5432)

This fixed all my problems on Mac OSX Mojave 10.14.1.

answered Dec 18 '18 at 01:03

Aaron Becker

593
1
6
10

I regret running this command, I killed my entire postgress and now it cannot start. – Billions.A.Joel Mar 29 '22 at 07:54
This is dangerous. You don't want to kill the base system postgres. – Reverse Engineered Jun 28 '22 at 01:49

score 18 · Answer 6 · answered Mar 07 '15 at 16:51

I have tried just about every solution to this problem that is out there. For me it always happens when my MacBook's battery dies, even if the computer is already sleeping. I was poking around in ~/Library/Application\ Support/Postgres/var-9.4 and I discovered another postmaster.pid file that I had not seen before. I deleted it, and now everything is back up and running! I am running the Postgres.app version, not the brew version.

Steps I took:

Make sure postgres.app is not running.
Run rm ~/Library/Application\ Support/Postgres/var-9.$X/postmaster.pid
If you don't have a var-9.$X directory, just run rm ~/Library/Application\ Support/Postgres/postmaster.pid
Restart postgres.app
Get back to developing cool stuff.

I had the same issue. This answer provided solution on my end. The key part is to make sure the system wouldn't be able to read that `postmaster.pid` file. Removing it via `rm` or just renaming it via `mv` would do. I prefer renaming it to `postmaster.pid.problem-cause` though. People have different installation setups so the best way to find that nasty file is to check the PosgtreSQL logs. — Abel Callejo, May 02 '19 at 01:05

score 17 · Accepted Answer · answered Oct 18 '12 at 01:49

17

You my have another instance of postgres running, that is the only application interested in this port. You should use netstat and ps to determine this. Then stop the instance and uninstall it, you mayhave installed enterpiseDb for example, that is what I did.

answered Oct 18 '12 at 01:49

Paul McLachlan

194
1
3

In my case this just happens randomly after some days of installing PostGreSQL on MacOSX. I then reinstall and everything works and then again after few days, it shows the same error. I want to make clear that i never change my root password for PostGreSQL. Please help – zulkarnain shah Sep 13 '17 at 07:13

score 10 · Answer 8 · answered May 26 '20 at 11:07

10

I generally face this issue on my mac, and this fixes it for me always

rm /usr/local/var/postgres/postmaster.pid
brew services restart postgresql

Hope this is helpful

answered May 26 '20 at 11:07

Utsav Preet

248
3
9

score 9 · Answer 9 · answered Dec 26 '20 at 05:35

9

This is what worked for me: $ sudo pkill -u postgres

Props to this resource: https://github.com/PostgresApp/PostgresApp/issues/197#issuecomment-474534056

answered Dec 26 '20 at 05:35

David Shi

186
3
3

ksnabb · Answer 10 · 2013-12-03T20:34:17.683

8

I had a similar problem where I could not connect to the Postgres.app even though the app itself said that it is running on port 5432.

I am not sure why, but even when I quit the app and checked that no postgres processes was running with ps -a. these files existed:

/tmp/.s.PGSQL.5432

and

/tmp/.s.PGSQL.5432.lock

My solution was to delete these files and then start the postgres.app again.

edited Dec 03 '13 at 20:34

answered Oct 28 '13 at 07:44

ksnabb

183
2
8

1

This solved the problem for me. Not sure why this answer was downvoted by others unless it's for some reason dangerous to delete `/tmp` files. – jurassic Nov 05 '13 at 07:08
2

I got excited at first, when seeing those files via `su` (superuser) and deleting them, but no joy – ckhatton Mar 12 '14 at 17:14
I tried "brew services stop postgresql" then deleted the /tmp/.s.PGSQL.5432.lock – HibernatedGuy Feb 09 '16 at 06:17

score 5 · Answer 11 · answered Apr 11 '16 at 20:17

5

It can also happen that the PID is taken. This ocurred to me when the Computer suffered an unexpected reboot.

If so, you must go to:

˜/Library/Application Support/Postgres/var-9.4

You can check that path in Postgres.app Preferences dialog.

And then, just remove the pid file

sudo rm postmaster.pid

And the server starts up right away.

answered Apr 11 '16 at 20:17

Lomefin

1,173
12
43

No such file in that folder. I am using PostGres 9.6 and couldnt find any file ending in that extension. – zulkarnain shah Sep 13 '17 at 07:16
Then your pid is not taken and something else is going on. – Lomefin Sep 25 '17 at 23:40

AdjunctProfessorFalcon · Answer 12 · 2017-07-06T19:17:46.600

When trying to open the Postgres app was getting that same error regarding post 5432 (on Mac OSX 10.10.5)

I did:

$ lsof -i | grep LISTEN

Saw which PID was running postgres at that port, did: $ killall {pid} and then $ brew uninstall postgres

After that, restarted my Mac, and ran:

$ lsof -i | grep LISTEN again just to make sure. Saw no postgres running anywhere and was able to open the Postgres app without getting that warning.

Then I reinstalled postgres with $ brew install postgres

ANOTHER OPTION:

(I also tried this one time when the above approach didn't work and I could not kill any of the PID)

$ ps auxw | grep post

Saw a slew postgres processes, which I then sudo kill <PID>

Everything working fine now.

Why do you want an postgres installed by homebrew and from the Postgres.app? — acidjunk, Jan 31 '19 at 07:22

score 3 · Answer 13 · answered Mar 21 '22 at 10:52

3

For those using mac, this code worked for me like charm.

sudo pkill -u postgres

answered Mar 21 '22 at 10:52

Billions.A.Joel

207
1
9

score 2 · Answer 14 · answered Aug 20 '12 at 02:07

At a guess, something else had taken port 5432 so the app chose to run on 5433 instead.

Why not just connect to Pg on port 5433, if that's where it's running? You have a /tmp/.s.PGSQL.5432 socket file, so you can connect with psql -p 5433 for UNIX domain socket connections. It'll also be listening on the same port with TCP/IP, so you can use psql -h localhost -p 5433 for TCP/IP and have your apps connect to port 5433 instead of 5432.

Change your .psqlrc to set the new port as default and you can forget it isn't on the default port.

Every option is worth exploring at this point, but this did not work for me. I am planning on re-installing the operating system at this point because I have not been able to remedy. — Brian McDonough, Aug 20 '12 at 17:43

score 2 · Answer 15 · answered Aug 16 '13 at 14:45

I had similar problem when trying to use postgresql with rails. Updating my Gemfile to use new version of gem pg solve this problem for me. (gem pg version 0.16.0 works). In the Gemfile use:

gem 'pg', '0.16.0'

then run the following to update the gem

bundle install --without production
bundle update
bundle install

score 1 · Answer 16 · answered Oct 10 '12 at 02:28

Mine failed to start up suddenly, and when I checked Console.app I saw:

com.heroku.postgres-service: 
FATAL:  could not create shared memory segment: Cannot allocate memory
DETAIL:  Failed system call was shmget(key=5432001, size=3874816, 03600).
HINT:  This error usually means that PostgreSQL's request for a shared memory segment exceeded available memory or swap space, or exceeded your kernel's SHMALL parameter.  You can either reduce the request size or reconfigure the kernel with larger SHMALL.  To reduce the request size (currently 3874816 bytes), reduce PostgreSQL's shared memory usage, perhaps by reducing shared_buffers or max_connections.
The PostgreSQL documentation contains more information about shared memory configuration.

It turns out Postgres wouldn't start up because I had Wireshark (and X11) running. It worked fine after I quit Wireshark.

Good luck!

I also had postgres.app fail to start up properly while X11 was running, killing X11 and restarting the app again fixed it. Mysterious problem. — Pathogen, Apr 20 '13 at 01:30

score 1 · Answer 17 · answered Jul 09 '13 at 12:54

I resolved this problem by

Identifying what was running on port 5432 by using "netstat" in the CL, which was postgreSQL not Postgres
I located the directory that contained postgreSQL, which was root/Library/PostgreSQL
I ensured an instants of the application wasn't running via Activity Monitor
Then I deleted the folder and rebooted! Everything was fine!

score 1 · Answer 18 · answered Aug 10 '13 at 21:15

1

Somehow I totally forgot that this socket file will be hidden because of the dot. Make sure you use ls -A /tmp/.s.PGSQL.5432 if you are checking to see if the socket is actually there.

answered Aug 10 '13 at 21:15

David Winiecki

4,093
2
37
39

score 1 · Answer 19 · answered Feb 26 '14 at 09:52

You most likely has a PostgreSQL installed, deleted it and installed it again. PostgreSQL typically used port 5432 but if not available, increases to the next available one, in this case 5433. So, you probably chose this port on your second install.

I think you should check file:

/etc/services

and adjust rows below for your expected port number:

postgresql 5432/udp # PostgreSQL Database

postgresql 5432/tcp # PostgreSQL Database

After this you should restart your computer (simplest way).

Kris Utter · Answer 20 · 2016-04-20T17:53:57.750

1

I had the same issues:

psql: could not connect to server: No such file or directory
    Is the server running locally and accepting
    connections on Unix domain socket "/tmp/.s.PGSQL.5432"?

In my case it was a corrupt host file on my mac. I was missing:

127.0.0.1   localhost

A quick easy way to check is to open terminal and type:

ping localhost

or:

scutil -r localhost

More info here Hope it helps.

edited Apr 20 '16 at 17:53

answered Nov 22 '14 at 03:43

Kris Utter

129
2
5

Daniel Golembiewski · Answer 21 · 2018-11-20T14:33:39.647

1

Netstat, ps aux, etc ... none showed 5432 in use. Checked /Library. Found PG9.6 old install still there. Did rm -rf and bang. Version 11 works fine.

edited Nov 20 '18 at 14:33

answered Nov 19 '18 at 12:57

Daniel Golembiewski

39
7

score 1 · Answer 22 · answered Jul 18 '21 at 11:55

I was losing my mind over this problem! i kept running

lsof -i | grep 5432

and nothing was showing up! finally i ran it using sudo and a potgres client showed up. so if anyone else has tried the lsof and went nowhere, try it with sudo.

sudo lsof -i | grep 5432

and then

sudo kill <the_pid_from_postgresql_found_from_lsof>

score 0 · Answer 23 · answered Sep 23 '13 at 15:12

The same problem just happened to me. I had Postgres.app 9.2.4.1 running since a while. When I updated Mac OS X to 10.8.5, after the mandatory reboot it was not working anymore. I tried several things, including updating to 9.2.4.3 and neither after another reboot it was working.

I had to open the file /Users/$USER/Library/Application Support/Postgres/var/postgresql.conf and to:

uncomment line unix_socket_permissions = 0777
uncomment and adapt line unix_socket_directory = '/tmp'

After a restart, Postgres.app was running like a charm.

score 0 · Answer 24 · answered Sep 30 '14 at 16:29

A file named "postmaster.pid" (in my $PGDATA directory) were preventing postgresql to start. It was a zombie file, placed there 10 days ago and when I brutally shut down the computer (pulling the plug, literally), no clean up process had the opportunity to remove that file.

score 0 · Answer 25 · answered Aug 21 '19 at 07:17

0

beside all the helpful answers, you might also want to take a look at official page instructions : postgressapp.com/documentation/remove.html

answered Aug 21 '19 at 07:17

narcis dpr

939
2
12
32

score 0 · Answer 26 · answered Dec 16 '20 at 17:06

0

I had this same issue and mine was caused by some configuration issues I was having. Clearing the user configurations and reinstalling postgres in brew worked for me!

brew uninstall postgresql
rm -rf /usr/local/var/postgres

answered Dec 16 '20 at 17:06

Rharris389

89
8

This was the issue in my case, Postgres was silently running from terminal and the installer app is unable to connect to the port 5432 – Srikar Reddy Jan 12 '22 at 05:15

score 0 · Answer 27 · answered Feb 21 '23 at 06:33

0

I ran the following:

psql --version then brew services stop postgresql@<VERSION>

finally, start your Postgres app on your Mac.

answered Feb 21 '23 at 06:33

Richard

118
1
6

Viand Direct · Answer 28 · 2023-07-15T12:43:30.110

0

I also had this problem,

TO solve, goto /tmp

The files you are looking for might be hidden so CMD+SHIFT+. will make them visible then delete

these files: /tmp/.s.PGSQL.5432

enter image description here /tmp/.s.PGSQL.5432.lock

you might need to restart your computer to get it working

edited Jul 15 '23 at 12:43

answered Jul 15 '23 at 11:44

Viand Direct

489
1
4
4

Postgres.app Could not start on port 5432

28 Answers28

Linked