Related
I have downloaded Portainer onto my server and created a PostgreSQL database in a container there.
Today I could no longer get access to the database. The log shows a message that there is a version problem.
I already read into some similar issues like Postgres container crashes with `database files are incompatible with server` after container's image has been updated to the latest one and Postgres container crashes with `database files are incompatible with server` after container's image has been updated to the latest one
and the solutions brew postgresql-upgrade-database did not work.
What can I do?
LOG
2021-10-03 [1] FATAL: database files are incompatible with server
2021-10-03 [1] DETAIL: The data directory was initialized by PostgreSQL version 13, which is not compatible with this version 14.0 (Debian 14.0-1.pgdg110+1).
PostgreSQL Database directory appears to contain a database; Skipping initialization
I also found this https://www.postgresql.org/docs/14/upgrading.html but the commands didn't work. Do I need to do this in the container somehow, or what commands will work to keep it running in the container?
You need to update the data file to the new format with this command:
$ brew postgresql-upgrade-database
I resolved it by
Removing the postgres image
Remove the volume
Pull the image again
Assuming you know the docker commands for above step.
On top of the answer suggesting removing all volumes/images/containers - if you have a shared volume for the DB in docker-compose.yml, like:
db:
image: postgres:14.1-alpine
volumes:
- ./tmp/db:/var/lib/postgresql/data
You will also need to remove the postgres mapped volume data which lives in ./tmp/db. Otherwise those conflicting files would still be there.
If you don't care about the data - you use the database for development purposes and you'll be able to re-create the db easily, just run rm -r ./tmp/db. Than you can just docker-compose up and re-create your database.
In case you care about the data, use pg_dumpall to dump you data before removing the files and restore after you run docker-compose up and your postgres service is ready again.
had a problem after running
brew update
brew upgrade
Brew upgraded postgresql to 14 which gave me
psql: could not connect to server: No such file or directory Is the server running locally and accepting connections on Unix domain socket "/var/run/postgresql/.s.PGSQL.5432"?
I have also seen the same error you have by inspecting
tail /usr/local/var/postgres/server.log
I have downgraded to version 13 running
brew uninstall postgresql
brew install postgresql#13
echo 'export PATH="/usr/local/opt/postgresql#13/bin:$PATH"' >> ~/.zshrc
and the command I've used to start the DB again is
pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start
I think because of the difference postgresql13 with 14 in initialization.
You can make a backup of your database before the new version, delete it and migrate database to the new version
for example in django:
dump: ./manage.py dumpdata -o mydata.json
load: ./manage.py loaddata mydata.json
django: Of course, if you have important information, do this separately for each model and pay attention to the dependencies when loading.
In docker-compose.yml you could change
db-data:/var/lib/postgresql/data:rw
to this
db-data:/var/lib/postgresql#14/data:rw
Delete image and run
docker-compose up -d
one more time
I was also facing the same issue with postgres in keycloak.
Downgrading the version to 13 resolved my issue.
for me the database directory was in /tmp/db and since no important data was there I remove all and everything worked well ...
but if your data is important then read these:
https://www.postgresql.org/docs/13/upgrading.html
https://www.postgresql.org/docs/13/pgupgrade.html
https://www.kostolansky.sk/posts/upgrading-to-postgresql-14/
I resolved it by
Removing the postgres container
docker rm "container name"
Removing the postgres image
docker rmi "image name"
docker-compose stop
docker-compose down
list all volumes
docker volume ls
remove volume
docker volume rm "volume name"
Pull the image again
Assuming you know the docker commands for above step.
I had the same issue after home brew update on my mac - Install #14 of postgresql
This is what help to me:
Install old binaries v13
brew install postgresql#13
Backup Your old DB folder
Initdb with new name
now use pg_upgrade like this:
pg_upgrade --old-datadir [Your old DB folder] --new-datadir [Your new
DB folder] --old-bindir [link to old bin (v13)]
for example:
pg_upgrade --old-datadir PSQ-data --new-datadir PSQ-data_new
--old-bindir /usr/local/Cellar/postgresql#13/13.9/bin
Source for old bin You will have after install #13, just look what you receive on terminal after install "Summary".
After this operation just start sever with new DB
The error 5: Input/output Error: Failure while executing; /bin/launchctl bootstrap gui/502 /Users/Andrea/Library/LaunchAgents/homebrew.mxcl.postgresql.plist exited with 5 is occurring when typing in the command "brew services start postgresql".
I am trying to create a database using homebrew and postgresql. If I type the command "brew services stop postgresql" and then start it back up the error does not occur. But will reoccur on the initial start of postgresql.
After, I stop postgresql and restart it using homebrew I type in "createdb 'test'. This error appears:
"error: could not connect to database template1: FATAL: password authentication failed for user "Andrea". Initially I tried to primarily work on the password authentication with no luck.
Here are the steps I have taken:
I have changed the method column to trust in my hba_pg files.
I have tried the command ALTER in the terminal to change the password.
I created a root login on my computer, where I tried to access the file and use the root's password to create the database.
The command chpass to change my user's password.
Help would be much appreciated. Thank you!
This is what I did to resolve on my Mac.
Looked at the postgres.log cat /usr/local/var/log/postgres.log | tail -100
Found the following error on my logs FATAL: lock file "postmaster.pid" already exists
Fix above PID issue and restart the server.
rm /usr/local/var/postgres/postmaster.pid
brew services restart postgresql
Sometimes this can be solved by simply running
brew services restart postgresql
instead of
brew services start postgresql
It happens when a previous postgres process doesn't terminate properly (maybe because you killed a psql process in activity monitor or otherwise). Restarting appears to properly unlock/delete the postmaster.pid file where "start" doesn't.
On the m1 mac I had the same issue and had to remove the postmaster.pid file as well. For m1 macs, the file is at /opt/homebrew/var/postgres/postmaster.pid, so I ran
/opt/homebrew/var/postgres/postmaster.pid
and I could brew services start postgresql successfully.
The other answers didn't apply because I didn't have the pid file (or many other files in those directories). The same error was caused by a bad install, most likely because I had previously installed libpq to get the psql CLI in isolation previously as per: Correct way to install psql without full Postgres on macOS?.
Solution, clean install:
> brew uninstall postgresql
> brew uninstall libpq
> brew install postgresql
> brew services run postgresql
To build on what #gajen-sunthara said, I made sure to upgrade my postgresql installation:
brew services upgrade postgresql
Then removed the postmaster.pid:
rm /usr/local/var/postgres/postmaster.pid
And restarted postgresql:
brew services restart postgresql
And that should get it going again :)
I got the same error:
% brew services start yabai
Bootstrap failed: 5: Input/output error
Try re-running the command as root for richer errors.
Error: Failure while executing; `/bin/launchctl bootstrap gui/503 /Users/myuser/Library/LaunchAgents/
homebrew.mxcl.yabai.plist` exited with 5.
I have an extra user who manages brew called brew.
After trying to run a brew service with sudo, this was a solution (on Apple silicon):
% sudo brew services stop yabai
% sudo chmod -R g+w /opt/homebrew
% sudo chown -R brew /opt/homebrew
% brew services start yabai
I know this has already been answered, but for anyone seeing this in the future, here is some help.
This often happens because it wasn't properly stopped when it was used at a different time, and the version you are using now has a conflict with what you were running previously. There are a few ways to fix it.
Try stoping it with brew services stop postgresql
Start it with brew services start postgresql
You can also just do brew services restart postgresql and not do step 1 and 2. Hope this helps someone!
I'm trying to run psql on my Vagrant machine, but I get this error:
psql: could not connect to server: No such file or directory
Is the server running locally and accepting connections on
Unix domain socket "/var/run/postgresql/.s.PGSQL.5432"?
Note: Vagrant 1.9.2
Box: ubuntu/trusty64, https://atlas.hashicorp.com/ubuntu/boxes/trusty64
EDIT
Commands I've used in order to install and run postgres:
sudo apt-get update
sudo apt-get install postgresql
sudo su postgres
psql -d postgres -U postgres
I've had this same issue, related to the configuration of my pg_hba.conf file (located in /etc/postgresql/9.6/main). Please note that 9.6 is the postgresql version I am using.
The error itself is related to a misconfiguration of postgresql, which causes the server to crash before it starts.
I would suggest following these instructions:
Certify that postgresql service is running, using sudo service postgresql start
Run pg_lsclusters from your terminal
Check what is the cluster you are running, the output should be something like:
Version - Cluster Port Status Owner Data directory
9.6 ------- main -- 5432 online postgres /var/lib/postgresql/9.6/main
Disregard the '---' signs, as they are being used there only for alignment.
The important information are the version and the cluster. You can also check whether the server is running or not in the status column.
Copy the info from the version and the cluster, and use like so:
pg_ctlcluster <version> <cluster> start, so in my case, using version 9.6 and cluster 'main', it would be pg_ctlcluster 9.6 main start
If something is wrong, then postgresql will generate a log, that can be accessed on /var/log/postgresql/postgresql-<version>-main.log, so in my case, the full command would be sudo nano /var/log/postgresql/postgresql-9.6-main.log.
The output should show what is the error.
2017-07-13 16:53:04 BRT [32176-1] LOG: invalid authentication method "all"
2017-07-13 16:53:04 BRT [32176-2] CONTEXT: line 90 of configuration file "/etc/postgresql/9.5/main/pg_hba.conf"
2017-07-13 16:53:04 BRT [32176-3] FATAL: could not load pg_hba.conf
Fix the errors and restart postgresql service through sudo service postgresql restart and it should be fine.
I have searched a lot to find this, credit goes to this post.
Best of luck!
I had the same issue but non of the answers here helped.
How I fixed it (mac)
Try to start postgresql with pg_ctl -D /usr/local/var/postgres start
Look for the Error Message that says something like FATAL: could not open directory "pg_tblspc": No such file or directory.
Create that missing directory mkdir /usr/local/var/postgres/pg_tblspc
Repeat from step one until you created all missing directories
When done and then trying to start postgresql again it might say FATAL: lock file "postmaster.pid" already exists
Delete postmaster.pid: rm /usr/local/var/postgres/postmaster.pid
Start postgres with: pg_ctl -D /usr/local/var/postgres start
Done ✨
These two steps solved it for me on Mac:
rm /usr/local/var/postgres/postmaster.pid
brew services restart postgresql
For M1 Macs:
rm /opt/homebrew/var/postgres/postmaster.pid
brew services restart postgresql
In case you face this issue (reported by #luckyguy73): psql: FATAL: database "postgresql" does not exist
You can run
brew postgresql-upgrade-database
to fix it.
I am just posting this for anyone who is feeling lost and hopeless as I did when I found this question. It seems that sometimes by editing some psotgresql-related config files, one can accidentally change the permissions of the file:
Note how pg_hba.conf belongs to root, and users cannot even read it. This causes postgres to not be able to open this file and therefore not be able to start the server, throwing the error seen in the original question.
By running
sudo chmod +r pg_hba.conf
I was able to make this file once again accessible to the postgres user and then after running
sudo service postgresql start
Was able to get the server running again.
WARNING: This will remove the database
Use command:
rm -rf /usr/local/var/postgres && initdb /usr/local/var/postgres -E utf8
WARNING: This will remove the database
Within zsh:
rm -rf /usr/local/var/postgres && initdb /usr/local/var/postgres -E utf8
This is the only thing that worked for me after countless hours trouble shooting.
Does the /etc/postgresql/9.6/main/postgresql.conf show that port being assigned? On my default Xubuntu Linux install, mine showed port = 5433 for some reason as best as I can remember, but I did comment out the line in that same file that said listen_addresses = 'localhost' and uncommented the line listen_addresses = '*'. So maybe start and check there. Hope that helps.
This works for me:
pg_ctl -D /usr/local/var/postgresql#9.6 stop;
brew services stop postgresql#9.6;
brew services start postgresql#9.6;
I was able to solve the issue by running:
sudo systemctl start postgresql#9.5-main
In my case Postgres was managed through Homebrew Services (i.e. started via brew services start postgresql#10 Terminal command for the Postgres 10 that I use), and for that setup I had to discover a couple of essential steps to do before I could apply any advice in this thread. So I want to share just that piece as it may help someone who has the same setup.
NOTE: all the commands below are to be run in Terminal.
To give a quick background: After upgrading to macOS Big Sur I discovered that Postgres wasn't working and running psql results in the error mentioned in the original question above. I tried to start Postgres (via the brew services start postgresql#10 command), this resulted in a message Service postgresql#10 already started. If I tried to restart it (via the brew services restart postgresql#10) I got a message that it was stopped and then started successfully. But! This was a misleading message, and I spent quite some time searching for config issues etc. before discovering that the service was not started successfully in reality.
So, the way to investigate this is:
Make sure the service is started by running the brew services start postgresql#10 (the latter argument may be different depending on what your Homebrew package name is e.g. postgresql#12 or plain postgresql).
Run brew services list. This is the command that gives you the true state of the service. In my case it said that Postgres' status is error:
Name Status User Plist
postgresql#10 error Denis /Users/Denis/Library/LaunchAgents/homebrew.mxcl.postgresql#10.plist
redis started Denis /Users/Denis/Library/LaunchAgents/homebrew.mxcl.redis.plist
To investigate further open the config shown in the same command output in Plist column (I used nano /Users/Denis/Library/LaunchAgents/homebrew.mxcl.postgresql#10.plist to check it).
In the config look for the StandardErrorPath key, and open the file located in the value of that key, i.e. in the <string> tag following the key. In my case it was /usr/local/var/log/postgresql#10.log.
Open that log and check the latest error (I used nano /usr/local/var/log/postgresql#10.log and then Alt+/ to go to the end of the file).
Voila. That is the real error to investigate, which you can then look for in the previous answers or google for. I'm not covering the rest here, as the goal of this answer is to show how to find the real error if you use Homebrew Services to run Postgres. (In my case it was the lock file "postmaster.pid" already exists already covered in the previous answers, plus the path to check right in the error message, in my case /usr/local/var/postgresql#10).
In my case it was the lockfile postmaster.id that was not deleted properly during the last system crash that caused the issue. Deleting it with sudo rm /usr/local/var/postgres/postmaster.pid and restarting Postgres solved the problem.
I recommend you should clarify port that postgres.
In my case I didn't know which port postgres was running on.
lsof -i | grep 'post'
then you can know which port is listening.
psql -U postgres -p "port_in_use"
with port option, might be answer. you can use psql.
If non of the above answers are not working for you, then please try this one,
Many people have mentioned many solutions to this problem! But all of them forgot that, the same problem will arise when your disk don't have enough space or the space you are assigned for postgres is full
Check your system storage, if its full free up some space! then restart your postgres by sudo service postgresql restart or do a stop and start sudo service posgresql stop then sudo service postgresql start
This will solve the issue, it solved for me
I occasionally have the same issue but mostly after macOS upgrades. Shutting down and migrating to the new version usually fixes it for me(make changes according to your version). So first upgrade your postgresql
brew services stop postgresql#12
brew services start postgresql#12
brew postgresql-upgrade-database
This is mostly a temporary fix but since I couldn't find a better solution this works for me.
Update: If the issue says that another postmaster is running then try removing it from that location(your postmaster.pid location will be displayed to you)
rm /usr/local/var/postgres/postmaster.pid
Open your database manager and execute this script
update pg_database set datallowconn = 'true' where datname = 'your_database_name';
I had the same error when I create the SQL db in a VM. I had changed the default value of /etc/postgresql/9.3/main/postgresql.conf shared_buffers = 200MB to 75% of my total RAM. Well, I forgot to actually allocate that RAM in the VM. When I gave the command to make a new database, I received the same error.
Powered off, gave the baby its bottle (RAM) and presto, it worked.
The same thing happened to me as I had changed something in the /etc/hosts file. After changing it back to 127.0.0.1 localhost it worked for me.
just reinstall your pgsql with direct version sudo apt-get install postgresql-9.5 (u must remove the package before install new one)
I had similar problems just a while ago. After trying more than 5 suggestions I decided to go back to the basics and start from the beginning. Which meant removing my postgresql installation and following this guide upon re-installing postgresql. https://help.ubuntu.com/lts/serverguide/postgresql.html
Ubuntu 20
This Problem happened to me, as ubuntu pre-installed version of Postgresql-9.6 server was always down and after trying all the above answers it didn't start.
Solution:
I installed another version of Postgresql which is postgresql-13, using this command: sudo apt install postgresql it will install the latest version of postgresql.
I see if the server is online or down using this command: pg_lsclusters if the new version of postgresql is online, we will proceed to remove the old version of postgresql.
we will see all packages that are installed related to postgresql, using this command: dpkg -l | grep postgresql
Remove the old version, which is here postgresql-9.6. Using this command:
sudo apt-get --purge remove postgresql-9.6 postgresql-client-9.6 replace 9.6 with your old version number. Final remaining packages related to the latest Version 13:
Restart your postgresql latest version server, which is here postgresql-13. Using this command: sudo systemctl restart postgresql#13-main replace 13 in the command with your latest version number.
Now, if you try psql command you will get an error related to your user, as in the image:
To Remove the above error, The installation procedure created a user account called postgres that is associated with the default Postgres role, to switch over to the postgres account use this command: sudo -u postgres psql this command will log you into the interactive Postgres session. You can also set your password for this user using this command \password postgres.
Then change the Port to the deafult port of postgresql, which is 5432 as all application will try to connect to postgresql using this port by default, using this command: sudo nano /etc/postgresql/13/main/postgresql.conf, it will open postgresql configuration file, then search for port and change it to 5432. After that you need to restart the server using this command sudo systemctl restart postgresql#13-main. Note, Replace 13 in the command with your latest version.
If you want to create your own User/Role, use this command: sudo -u postgres createuser --interactive. The script will prompt you with some choices, as in the image and based on your responses, it will execute the correct Postgres commands to create a user to your specifications.
Tutorial: For more information on postgresql related commands
I couldn't connect using the psql command and kept getting the error Cannot connect to Server: No such file or directory.
Step 1: Check the status of the Postgres cluster
$ pg_lsclusters
Step 2: Restart the Postgres cluster
$ sudo pg_ctlcluster 12 main start
Make sure to replace 12 with your version of Postgres
Step 3: Check again and connect
$ pg_lsclusters
$ sudo -i -u postgres
$ psql
I got this error when I restored my database from last pg_basebackup backup file. After that when I tried to connect database(psql), I was getting the same error. The error was resolved, when I updated pg_hba.conf file and wherever "peer" authentication was there I replaced that with "md5" and then restarted postgres services. After that, the problem was resolved.
This error happened to me after my mac mini got un-plugged (so forced shutdown), and all I had to do to fix it was restart
I have the same issue with postgres 11 on my mac. I get this error every time after restart
psql: could not connect to server: No such file or directory
Is the server running locally and accepting connections on
Unix domain socket "/var/run/postgresql/.s.PGSQL.5432"?
As a temporary fix I do
brew services stop postgresql#11
brew services start postgresql#11
My problem happened after a brew update so I've ran
pg_ctl -D /usr/local/var/postgres start
and I've got this result:
FATAL: database files are incompatible with server 2021-07-07 13:27:21.692 CEST [70896] DETAIL: The data directory was initialized by PostgreSQL version 12, which is not compatible with this version 13.2. stopped waiting
I've ran
brew postgresql-upgrade-database
FATAL: could not load server certificate file "/etc/ssl/certs/ssl-cert-snakeoil.pem": No such file or directory
LOG: database system is shut down
pg_ctl: could not start server
I have a missing ssl-cert-snakeoil.pem file so i created it using make-ssl-cert generate-default-snakeoil --force-overwrite And it worked fine.
In my case, I had to run journalctl -xe, and it showed that my disk was full. I then deleted some .gz items from /var/log and I could again restart the postgresql.
I'm on Kali Linux. I had to remove the brew version of postgresql with
brew uninstall postgresql
sudo -u postgres psql got me into root postgres
Simply running these commands from the installation steps in the official PostgreSQL docs worked for me (I'm on Fedora 33):
# Optionally initialize the database and enable automatic start:
sudo /usr/pgsql-13/bin/postgresql-13-setup initdb
sudo systemctl enable postgresql-13
sudo systemctl start postgresql-13
RHEL Installation link
kali users pls do this
sudo service postgresql restart
I was following these steps to install postgresql and then odoo v9 :
https://www.odoo.com/documentation/9.0/setup/install.html
Basically it follows these steps outlined here:
https://www.odoo.com/forum/help-1/question/issue-while-installing-following-the-instructions-for-packaged-installers-rpm-93756
I get exactly the same situation. The first steps go through without issue but then the postgresql-setup initdb command cannot be found.
I have root access to the server and I know it's running centOS5.5. How can I troubleshoot this? I believe I installed the 9.4 version but when I do psql -V I get the server has v8.3.
I ran some commands on the server and here are the results:
postgresql-setup initdb
command cannot be found
postgres
root execution is not permitted, server must be restarted as underprivileged id...
initdb
no data directory specified, you must identify dir where datase system will reside
locate postgresql
i get an .rb, .xml and .yml files
locate psql
/usr/lib64 - a few libodbcpsql.so files
pqsl -V
8.1.23
I get command not found for all other commands in the list of commands to install odoo up until sudo yum config manager --add-repo etc...
sudo su postgres -c "psql -c 'show data_directory'"
could not change directory to 'the directory im currently in on my server'
Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
We don’t allow questions seeking recommendations for books, tools, software libraries, and more. You can edit the question so it can be answered with facts and citations.
Closed 1 year ago.
The community reviewed whether to reopen this question 1 year ago and left it closed:
Original close reason(s) were not resolved
Improve this question
I'm using the PostgreSQL database for my Ruby on Rails application (on Mac OS X 10.9).
Are there any detailed instructions on how to upgrade PostgreSQL database?
I'm afraid I will destroy the data in the database or mess it up.
Assuming you've used home-brew to install and upgrade Postgres, you can perform the following steps.
Stop current Postgres server:
launchctl unload ~/Library/LaunchAgents/homebrew.mxcl.postgresql.plist
Initialize a new 10.1 database:
initdb /usr/local/var/postgres10.1 -E utf8
run pg_upgrade (note: change bin version if you're upgrading from something other than below):
pg_upgrade -v \
-d /usr/local/var/postgres \
-D /usr/local/var/postgres10.1 \
-b /usr/local/Cellar/postgresql/9.6.5/bin/ \
-B /usr/local/Cellar/postgresql/10.1/bin/
-v to enable verbose internal logging
-d the old database cluster configuration directory
-D the new database cluster configuration directory
-b the old PostgreSQL executable directory
-B the new PostgreSQL executable directory
Move new data into place:
cd /usr/local/var
mv postgres postgres9.6
mv postgres10.1 postgres
Restart Postgres:
launchctl load ~/Library/LaunchAgents/homebrew.mxcl.postgresql.plist
Check /usr/local/var/postgres/server.log for details and to make sure the new server started properly.
Finally, re-install the rails pg gem
gem uninstall pg
gem install pg
I suggest you take some time to read the PostgreSQL documentation to understand exactly what you're doing in the above steps to minimize frustrations.
Despite all answers above, here goes my 5 cents.
It works on any OS and from any-to-any postgres version.
Stop any running postgres instance;
Install the new version and start it; Check if you can connect to the new version as well;
Change old version's postgresql.conf -> port from 5432 to 5433;
Start the old version postgres instance;
Open a terminal and cd to the new version bin folder;
Run pg_dumpall -p 5433 -U <username> | psql -p 5432 -U <username>
Stop old postgres running instance;
Here is the solution for Ubuntu users
First we have to stop postgresql
sudo /etc/init.d/postgresql stop
Create a new file called /etc/apt/sources.list.d/pgdg.list and add below line
deb http://apt.postgresql.org/pub/repos/apt/ utopic-pgdg main
Follow below commands
wget -q -O - https://www.postgresql.org/media/keys/ACCC4CF8.asc | sudo apt-key add -
sudo apt-get update
sudo apt-get install postgresql-9.4
sudo pg_dropcluster --stop 9.4 main
sudo /etc/init.d/postgresql start
Now we have everything, just need to upgrade it as below
sudo pg_upgradecluster 9.3 main
sudo pg_dropcluster 9.3 main
That's it. Mostly upgraded cluster will run on port number 5433. Check it with below command
sudo pg_lsclusters
If you are using homebrew and homebrew services, you can probably just do:
brew services stop postgresql
brew upgrade postgresql
brew postgresql-upgrade-database
brew services start postgresql
I think this might not work completely if you are using advanced postgres features, but it worked perfectly for me.
Update: This process is the same for upgrading 9.5 through at least 11.5; simply modify the commands to reflect versions 9.6 and 10, where 9.6 is the old version and 10 is the new version. Be sure to adjust the "old" and "new" directories accordingly, too.
I just upgraded PostgreSQL 9.5 to 9.6 on Ubuntu and thought I'd share my findings, as there are a couple of OS/package-specific nuances of which to be aware.
(I didn't want to have to dump and restore data manually, so several of the other answers here were not viable.)
In short, the process consists of installing the new version of PostgreSQL alongside the old version (e.g., 9.5 and 9.6), and then running the pg_upgrade binary, which is explained in (some) detail at https://www.postgresql.org/docs/9.6/static/pgupgrade.html .
The only "tricky" aspect of pg_upgrade is that failure to pass the correct value for an argument, or failure to be logged-in as the correct user or cd to the correct location before executing a command, may lead to cryptic error messages.
On Ubuntu (and probably Debian), provided you are using the "official" repo, deb http://apt.postgresql.org/pub/repos/apt/ xenial-pgdg main, and provided you haven't changed the default filesystem paths or runtime options, the following procedure should do the job.
Install the new version (note that we specify the 9.6, explicitly):
sudo apt install postgresql-9.6
Once installation succeeds, both versions will be running side-by-side, but on different ports. The installation output mentions this, at the bottom, but it's easy to overlook:
Creating new cluster 9.6/main ...
config /etc/postgresql/9.6/main
data /var/lib/postgresql/9.6/main
locale en_US.UTF-8
socket /var/run/postgresql
port 5433
Stop both server instances (this will stop both at the same time):
sudo systemctl stop postgresql
Switch to the dedicated PostgreSQL system user:
su postgres
Move into his home directory (failure to do this will cause errors):
cd ~
pg_upgrade requires the following inputs (pg_upgrade --help tells us this):
When you run pg_upgrade, you must provide the following information:
the data directory for the old cluster (-d DATADIR)
the data directory for the new cluster (-D DATADIR)
the "bin" directory for the old version (-b BINDIR)
the "bin" directory for the new version (-B BINDIR)
These inputs may be specified with "long names", to make them easier to visualize:
-b, --old-bindir=BINDIR old cluster executable directory
-B, --new-bindir=BINDIR new cluster executable directory
-d, --old-datadir=DATADIR old cluster data directory
-D, --new-datadir=DATADIR new cluster data directory
We must also pass the --new-options switch, because failure to do so results in the following:
connection to database failed: could not connect to server: No such file or directory
Is the server running locally and accepting
connections on Unix domain socket "/var/lib/postgresql/.s.PGSQL.50432"?
This occurs because the default configuration options are applied in the absence of this switch, which results in incorrect connection options being used, hence the socket error.
Execute the pg_upgrade command from the new PostgreSQL version:
/usr/lib/postgresql/9.6/bin/pg_upgrade --old-bindir=/usr/lib/postgresql/9.5/bin --new-bindir=/usr/lib/postgresql/9.6/bin --old-datadir=/var/lib/postgresql/9.5/main --new-datadir=/var/lib/postgresql/9.6/main --old-options=-cconfig_file=/etc/postgresql/9.5/main/postgresql.conf --new-options=-cconfig_file=/etc/postgresql/9.6/main/postgresql.conf
Logout of the dedicated system user account:
exit
The upgrade is now complete, but, the new instance will bind to port 5433 (the standard default is 5432), so keep this in mind if attempting to test the new instance before "cutting-over" to it.
Start the server as normal (again, this will start both the old and new instances):
systemctl start postgresql
If you want to make the new version the default, you will need to edit the effective configuration file, e.g., /etc/postgresql/9.6/main/postgresql.conf, and ensure that the port is defined as such:
port = 5432
If you do this, either change the old version's port number to 5433 at the same time (before starting the services), or, simply remove the old version (this will not remove your actual database content; you would need to use apt --purge remove postgresql-9.5 for that to happen):
apt remove postgresql-9.5
The above command will stop all instances, so you'll need to start the new instance one last time with:
systemctl start postgresql
As a final point of note, don't forget to consider pg_upgrade's good advice:
Upgrade Complete
----------------
Optimizer statistics are not transferred by pg_upgrade so,
once you start the new server, consider running:
./analyze_new_cluster.sh
Running this script will delete the old cluster's data files:
./delete_old_cluster.sh
The user manual covers this topic in depth. You can:
pg_upgrade in-place; or
pg_dump and pg_restore.
If in doubt, do it with dumps. Don't delete the old data directory, just keep it in case something goes wrong / you make a mistake; that way you can just go back to your unchanged 9.3 install.
For details, see the manual.
If you're stuck, post a detailed question explaining how you're stuck, where, and what you tried first. It depends a bit on how you installed PostgreSQL too, as there are several different "distributions" of PostgreSQL for OS X (unfortunately). So you'd need to provide that info.
Standing on the shoulders of the other poor creatures trodding through this muck, I was able to follow these steps to get back up and running after an upgrade to Yosemite:
Assuming you've used home-brew to install and upgrade Postgres, you can perform the following steps.
Stop current Postgres server:
launchctl unload ~/Library/LaunchAgents/homebrew.mxcl.postgresql.plist
Initialize a new 9.4 database:
initdb /usr/local/var/postgres9.4 -E utf8
Install postgres 9.3 (as it was no longer present on my machine):
brew install homebrew/versions/postgresql93
Add directories removed during Yosemite upgrade:
mkdir -p /usr/local/var/postgres/{pg_tblspc,pg_twophase,pg_stat_tmp}/touch /usr/local/var/postgres/{pg_tblspc,pg_twophase,pg_stat_tmp}/.keep
run pg_upgrade:
pg_upgrade -v -d /usr/local/var/postgres -D /usr/local/var/postgres9.4 -b /usr/local/Cellar/postgresql93/9.3.5/bin/ -B /usr/local/Cellar/postgresql/9.4.0/bin/
Move new data into place:
cd /usr/local/var
mv postgres postgres9.3
mv postgres9.4 postgres
Restart Postgres:
launchctl load ~/Library/LaunchAgents/homebrew.mxcl.postgresql.plist
Check /usr/local/var/postgres/server.log for details and to make sure the new server started properly.
Finally, re-install related libraries?
pip install --upgrade psycopg2
gem uninstall pg
gem install pg
Looks like the solution has been baked into Homebrew now:
$ brew info postgresql
...
==> Caveats
To migrate existing data from a previous major version of PostgreSQL run:
brew postgresql-upgrade-database
....
On Windows I kept facing different errors messages when trying to use pg_upgrade.
Saved a lot of time for me to just:
Backup DB
Uninstall all copies of PostgreSQL
Install 9.5
Restore DB
This did it for me.
https://gist.github.com/dideler/60c9ce184198666e5ab4
Short and to the point. I honestly don't aim to understand the guts of PostgreSQL, I want to get stuff done.
My solution for upgrading from Postgresql 11 to Postgresql 12 on Windows 10 is the following.
As a first remark you will need to be able stop and start the Postgresql service. You can do this by the following commands in Powershell.
Start:
pg_ctl start -D “d:\postgresql\11\data”
Stop:
pg_ctl stop -D “d:\postgresql\11\data”
Status:
pg_ctl status -D “d:\postgresql\11\data”
It would be wise to make a backup before doing the upgrade. The Postgresql 11 instance must be running. Then to copy the globals do
pg_dumpall -U postgres -g -f d:\bakup\postgresql\11\globals.sql
and then for each database
pg_dump -U postgres -Fc <database> > d:\backup\postgresql\11\<database>.fc
or
pg_dump -U postgres -Fc -d <database> -f d:\backup\postgresql\11\<database>.fc
If not already done install Postgresql 12 (as Postgresql 11 is also installed this will be on port 5433)
Then to do the upgrade as follows:
1) Stop Postgresql 11 service (see above)
2) Edit the postgresql.conf file in d:\postgresql\12\data and change port = 5433 to port = 5432
3) Edit the windows user environment path (windows start then type env) to point to Postgresql 12 instead of Postresql 11
4) Run upgrade by entering the following command.
pg_upgrade `
-b “c:\program files\postgresql\11\bin” `
-B “c:\program files\postgresql\12\bin” `
-d “d:\postgresql\11\data” `
-D “d:\postgresql\12\data” --username=postgres
(In powershell use backtick (or backquote) ` to continue the command on the next line)
5) and finally start the new Postgresql 12 service
pg_ctl start -D “d:\postgresql\12\data”
My solution was to do a combination of these two resources:
https://gist.github.com/tamoyal/2ea1fcdf99c819b4e07d
and
http://www.gab.lc/articles/migration_postgresql_9-3_to_9-4
The second one helped more then the first one. Also to not, don't follow the steps as is as some are not necessary.
Also, if you are not being able to backup the data via postgres console, you can use alternative approach, and backup it with pgAdmin 3 or some other program, like I did in my case.
Also, the link: https://help.ubuntu.com/stable/serverguide/postgresql.html
Helped to set the encrypted password and set md5 for authenticating the postgres user.
After all is done, to check the postgres server version run in terminal:
sudo -u postgres psql postgres
After entering the password run in postgres terminal:
SHOW SERVER_VERSION;
It will output something like:
server_version
----------------
9.4.5
For setting and starting postgres I have used command:
> sudo bash # root
> su postgres # postgres
> /etc/init.d/postgresql start
> /etc/init.d/postgresql stop
And then for restoring database from a file:
> psql -f /home/ubuntu_username/Backup_93.sql postgres
Or if doesn't work try with this one:
> pg_restore --verbose --clean --no-acl --no-owner -h localhost -U postgres -d name_of_database ~/your_file.dump
And if you are using Rails do a bundle exec rake db:migrate after pulling the code :)
For Mac via homebrew:
brew tap petere/postgresql,
brew install <formula> (eg: brew install petere/postgresql/postgresql-9.6)
Remove old Postgres:
brew unlink postgresql
brew link -f postgresql-9.6
If any error happen, don't forget to read and follow brew instruction in each step.
Check this out for more: https://github.com/petere/homebrew-postgresql
On Windows 10 since I had npm, I installed rimraf package. npm install rimraf -g
Backup all your databases one by one using command pg_dump -U $username --format=c --file=$mydatabase.sqlc $dbname
Then Installed Latest PostgreSQL Version i.e. 11.2 which prompted me to use port 5433 this time.
Followed by Uninstall of older versions of PostgreSQL mine was 10. Note the uninstaller may give a warning of not deleting folder C:\PostgreSQL\10\data. That's why we have the next step using rimraf to permanently delete the folder and it's sub-folders.
change into PostgreSQL install directory and ran the command rimraf 10. 10 is a directory name. Note use your older version of PostgreSQL i.e. 9.5 or something.
Now add C:\PostgreSQL\pg11\bin, C:\PostgreSQL\pg11\lib into the Windows environmental variables. Note my new installed version is 11 thus why I am using pg11.
Navigate to C:\PostgreSQL\data\pg11 then open postgresql.conf edit port = 5433 to port = 5432
That's it. Open cmd and type psql -U postgres
You can now restore all your backed databases one by one using the command pg_restore -U $username --dbname=$databasename $filename
I think this is best link for your solution to update postgres to 9.6
https://sandymadaan.wordpress.com/2017/02/21/upgrade-postgresql9-3-9-6-in-ubuntu-retaining-the-databases/