Using Manjaro, I freshly installed PostgreSQL with an user-defined data-directory. PostgreSQL is running fine, but I can't find the postgresql*.log(s), which I used to find in /var/log/postgresql on my former system. In ksystemlog I've seen an error, that I forced by
select * from non_exist_table;
but I'm wondering, if there is no longer the postgresql*.log.
Thank you!
Here is my postgresql.conf.
With logging_collector=off, the logs will just spew to the stderr of the postmaster process. That could have been redirected to a file, or just going to some tty someplace, or just be thrown away. You would have to look at the code which launches the database server.
At my work I was running a complex query. I cancelled it and went home yesterday. This morning in the back the query was impossible to be terminated, also with the 'terminate backend' functionality. A colleague of mine restarted the host machine where postgres is installed. After the machine restart, the postgres database sever would not start up.
In my log files I see the error:
'pg_ctl: this data directory appears to be running a pre-existing postmaster'
I am not sure how to handle this problem. I could try to fix it or try to extrapolate the data from the save files. What is the most logical step to take and do you know how to fix this?
Earlier it gave this error message :
2016-01-28 15:52:33 GMT FATAL: lock file "postmaster.pid" already exists
2016-01-28 15:52:33 GMT HINT: Is another postmaster (PID 2100) running in data directory "C:/PostgreSQL/9.1/data"?
UPDATE... I located the file postmaster.pid and deleted it. Now I am restarting the computer and hoping it will start.
UPDATE... It works now. I rebooted the computer and postgres just instantly started. Happy as a child but at the same time not fully satisfied because of the following forum: https://superuser.com/questions/553045/fatal-lock-file-postmaster-pid-already-exists . Here it is stated to NEVER delete the postmaster.pid because of possible data corruption. So because of that I will backup all databases I have in postgres now.
So if anyone can share some more light on my ICT adventure of today I would be very satisfied. That is why I will not state that this question is answered, since I have no idea what went wrong and perhaps will run into it again someday.
The explanation is pretty straightforward. PostgreSQL writes the process ID to a file called postmaster.pid — the presence of the file is supposed to indicate that the server is running. When the PostgreSQL shuts down cleanly, it removes the postmaster.pid file.
However, when your colleague restarted the host machine, the PostgreSQL server got killed without having had a chance to remove the postmaster.pid file. Therefore, when you tried to start PostgreSQL, the presence of the file made it look complain that the server was already running.
This answer provides more complete advice. In general, you should never delete postmaster.pid for no good reason, because it's supposed to help prevent two servers from running at once on the same data files. However, if you are certain that the process indicated by the postmaster.pid file is already dead, then by all means just delete the stale PID file manually.
In windows Delete all running postgres processes and start the service
I'm using:
CentOS 6.7
Zend.1.2.15 using oci8, using instantclient 11.2 x64
PHP 5.4.45
Trying to connect to an Oracle:
Oracle Database 10g Release 10.2.0.4.0
Already checked credentials, already tried to connet to other Oracle server but the error still the same:
12546 ORA-12546: TNS:permission denied *
Already gave 0777 permission to /usr/local/oracle/11.2/client64/ following some answer around but none of them solved.
Server does not have the IP in any sort of blacklist.
The same code runs in any other Windows PC with OracleInstantClient and no further configuration was needed on the server about giving an IP permissions to access or any permission at all.
Is there something more that I can check to solve this?
After hours searching I found this article: http://www.emsperformance.net/2013/12/20/connecting-to-oracle-database-from-php-with-oci8/ which is by far the best article explaining the installation process of LAMP, Oracle, Zend.
The setting I was missing is:
setsebool -P httpd_can_network_connect on
Edit: After a while this answer still looks useful to someone, worths to say that this happens because the machine has SELinux activated, usually in the enforcing mode. I would not recommend to disable SELinux, but setting it to permissive looks to ease the process of identifying some problems with it. You can check SELinux status by sestatus.
A full list of booleans on SELinux is available at: https://wiki.centos.org/TipsAndTricks/SelinuxBooleans
Overall information: https://wiki.centos.org/HowTos/SELinux
I have attempted to install PostgreSQL 9.4 and 8.4 multiple times and it is failing no matter what I have tried. I am attempting to install on Windows 7 SP1 x64. After each failed install I have uninstalled and deleted the installation folder to start fresh.
Each time I attempt the install I get an error pop up near the end of installation that says:
"failed to load sql modules into the database cluster".
Then another error pop up displays immediately after that says:
"Error running post install step. Installation may not complete correctly. Error reading C:/Program Files/PostgreSQL/9.4/postgresql.conf"
I have attempted installation with the following actions:
Always installed as administrator
Turned off all virus protection and windows firewall
Changed the installation directory to something other than the Program Files directory.
Changed the data directory to something other than the installation directory of postgres
None of the actions above have helped and I always receive the error. Any help that someone can provide would be greatly appreciated!
Encountered a very similar problem that OP is reporting today while installing Postgres 9.4.
It turns out that the password generator I was using has made a password that contains non-alphanumeric characters. ("^") I believe was the culprit in this case. Removing that allowed the installation of Postgres 9.4 (Windows 64 bit) to complete.
Very easy fix once you've found it, but the error message that comes up is not descriptive, so in this case I didn't notice at first what I was doing wrong.
I was getting this same error when trying to install PostgreSQL v9.4.4 on Windows 10 Pro. Starting with a solution hosted on Stack Exchange, I came up with the following steps that allowed the installer to run successfully:
1) Create a new user account, called postgres
2) Add the new account to the Administrators and Power Users groups
3) Restart the computer
NOTE: I added step #3, since step #4 didn't work without it
4) Run a command prompt as the postgres user, using the command:
runas /user:postgres cmd.exe
5) Run the installer from the postgres command window
6) Delete the postgres user account, as well as the user directory
NOTE: I added step #6, since the postgres account is not required after installation
What worked for me is, during the install, specifying a Postgre SQL data folder that's outside of any Windows user profile directory (C:\Users), such as C:\postgres-data.
My setup:
Win 10 Pro
PostgreSQL 9.5 RC1
I ran into this issue when I tried setting PostgreSQL's data directory somewhere under my user profile's directory, such us somewhere under "My Documents".
I tried Jeff G's solution and it didn't work at first. It worked only when I kept the data directory as the default (C:\Program Files\PostgreSQL\9.5\data). So then I tested further and tried setting the data directory to somewhere outside of any user profile directories (i.e. somewhere that isn't under C:\Users). For example, C:\postgres-data and this worked.
I then just tried using that directory with my default user, and not bothering with the postgres temp user as outlined in Jeff G's solution and that also worked. So in the end, it had to do with my data directory being somewhere under C:\Users. As long as it was outside of that, it worked.
None of these answers here helped me, finally I solved this problem by creating the folder before the installation (C:\PostgreSQL\data) and giving it full access for the group "Users".
Windows 7 x64, postgresql-9.5.2-1-windows-x64
Hope It's Work.
1.)Uninstall PostgreSQL
2.)Delete the postgres user if it still exists :
net user postgres /delete
3.) Create the postgres user with a password you can remember:
net user /add postgres
4.) Add the postgres user to the Administrators group:
net localgroup administrators postgres /add
5.) Add the postgres user to the Power Users group
net localgroup "power users" postgres /add
6.) Run a command window as the postgres user:
runas /user:postgres cmd.exe
7.) Change user postgres and install postgresql
8.)Back your and remove the postgres user from the Administrators group.
net localgroup administrators postgres /delete
I had the same error ("Failed to load SQL modules into the database cluster.") when installing on W2K12R2 using the EnterpriseDB installer linked to from the PostgreSQL Windows download page. I tried running the installer with admin privs, and using the postgres-user solution offered by Jeff G, but neither worked. Finally I tried the second installer, BigSQL, and that installed without issue.
When installing PostgreSQL, do not use the following symbols in your database's admin password: %, <, and >.
Bug report and problem solution here
Changed the data directory to something other than the installation directory of postgres
Make sure NETWORK SERVICE has read/write permissions on that folder, the installation was failing for me until I did this.
I had this issue too with 9.5 and got around it by:
Installing with the default data folder.
Making sure my new data folder had full control access for the "NETWORK SERVICE" account.
Then changing the default PGDATA folder as per this instruction: https://wiki.postgresql.org/wiki/Change_the_default_PGDATA_directory_on_Windows
I had the same problem, and noticed that some Postgres .bat files were opening in a text editor. Text editors opening files during installation isn't too uncommon, but I noticed these batch files were setting directory permissions. I then found that .bat files were set to open in a text editor (Notepad++) by default. I changed the default program for .bat files, uninstalled Postgres, re-ran the installer, and it worked perfectly. Hope this helps.
I had the same problem before. I solved this by installing Visual Studio C++ Express first.
I had the same experience as #gomisha. #Jeff G's solution did not work for me no matter which user ran the installer.
As long as the target data directory was anywhere under c:\users, the install or database cluster init would fail.
I was successful after creating c:\postgresql-data and ensuring that the postgres user had full access in the 'effective access' tool.
Note: I did not use control userpasswords2 to set up the postgres user; I did it through the traditional crappy Windows 10 user wizard that tries to make the user sign up for hotmail. I did run the install as postgres.
I was getting this issue with the BigSQL windows installer PostgreSQL-9.6.5-1-win64-bigsql.exe
The problem turned out to be that my system path was too long. After removing some uncessary junk from my System Path, and then reinstalling, the problem was gone.
One of the answers here is the only thing that worked on my machine:
I had the same error message when trying to install 9.2.4. My issue
was because even though the Win2k8 server had %SYSTEMROOT%\system32 as
part of the path, no programs could "see" anything in
C:\Windows\system32. The installer heavily uses icacls.exe during the
init routine. Since my path was screwed up, the installer bombed out.
Once I explicitly added C:\Windows\system32 to the SYSTEM's Path
environment variable and re-ran the installer as an admin, everything
worked fine.
Credits to the user.
Go to services.msc
Search for postgres service
Go to Log On tab
Select local account
Profit
The alternative solution for problems with PostgreSQL installer is setting this up with a Docker.
Install Docker Desktop
https://www.docker.com/products/docker-desktop/
Start installed Docker Desktop
In PowerShell, run:
docker pull postgres
In PowerShell, run:
docker run -d -p 5432:5432 --name postgres -e POSTGRES_PASSWORD=mypassword postgres
Now you can connect to the Postgres server for example with pgAdmin on the same machine, using:
host: localhost
port: 5432
maintenance database: postgres
username: postgres
password: mypassword
I think this would be better as individual comments above, but I don't have the required rep. I just spent days on this, with Postgres 10 on Win 10 Creators. My "answer" was to let Postgres put the data in its default install folder instead of my User folder (where it would be automatically backed-up).
1. This problem can happen with no reported install errors at all. On my very first try, I saw:
Failed to load SQL modules into the database cluster.
But that seemed to be due to:
Executing C:\Users\loren\AppData\Local\Temp\postgresql_installer_aee8e5a76f\vcredist_x64.exe /passive /norestart
Script exit code: 3010
MSI (s) (AC:FC) [14:21:35:341]: Product: Microsoft Visual C++ 2013 x86 Minimum Runtime - 12.0.40660.
Restart required. The installation or update for the product required a restart for all changes to take effect.
The restart was deferred to a later time.
Restarting Windows after the failure of the rest of the Postgres installation did not let Postgres run.
Once VC was in place, uninstalling and re-installing Postgres (still with its data in my User folder) completed with no reported errors. But every attempt to access it, by any means, still resulted in something like this:
C:\Program Files\PostgreSQL\10\bin>psql
psql: could not connect to server: Connection refused (0x0000274D/10061)
Is the server running on host "localhost" (::1) and accepting
TCP/IP connections on port 5432?
could not connect to server: Connection refused (0x0000274D/10061)
Is the server running on host "localhost" (127.0.0.1) and accepting
TCP/IP connections on port 5432?
The RawCap sniffer showed TCP SYN followed immediately by RST,ACK on localhost for both v4 and v6 for each try. Netstat did not see port 5432 in use at all. All Postgres runtime logs were totally empty. In Windows Services, I could manually start the postgresql-x64-10 service, but it immediately stopped itself, with the note that some services do that normally.
2. I tried moving the data directory via the Postgres wiki procedure. Apparently for Postgres 10 it is incomplete. It does not deal with:
Computer\HKEY_LOCAL_MACHINE\SOFTWARE\PostgreSQL\Installations\postgresql-x64-10\Data Directory
Computer\HKEY_LOCAL_MACHINE\SOFTWARE\PostgreSQL\Services\postgresql-x64-10\Data Directory
C:\Program Files\PostgreSQL\10\pg_env.bat
C:\Program Files\PostgreSQL\10\scripts\serverctl.vbs
But fixing every data file instance I could find did not let it run. And even after fixing all those, and uninstalling and re-installing again with the default location, the new pgAdmin4 somehow still remembered my original User data directory location!
3. I did not try giving NETWORK SERVICE read/write permissions on my User folder. Once the default install worked I wanted to get on with my original project... Maybe that would have worked.
When I try to connect to my Postgres.app db using dbext, I get the following error:
dbext:PostgreSQL requires a '$HOME/.pgpass' file in order to authenticate. This file is
missing. The binary 'psql' does not accept commandline passwords.
Other programs connect just fine by using a "local" connection. (Postgres.app runs with my userid.)
In vim :!which psql correctly prints /opt/local/bin/psql (which I have symlinked to the one in the Postgres.app bin directory). And Postgres.app is set up to use "local" authentication and there's no clear sense of where a pg_hba.conf file would go (there is no etc directory in the app bundle). Moreover, Postgres.app doesn't have anything in its documentation about changing access configuration.
I've tried using dbext's :DBPromptForBufferParameters directly, as well as #tpope's vim-rails plugin (which returns without comment from dbext setup via :Rdbext.
So what do I do to get dbext to connect using a "local" connection?
Note - I spent a LOT of time trying to figure this out without trying the obvious, thus the post even when I already have the answer. I'm also curious to see if anyone else has a different approach.
It turns out you can just make an empty ~/.pgpass file (restricting read-write permissions to your userid only to avoid warnings). This was counterintuitive for me (since there is in fact no password), but I suppose in retrospect it's obvious I should have tried it.
I'll point this out on the dbext issue tracker.