When I try to run pg_dump as the user "postgres",
$ sudo -u postgres pg_dump <pg_dump arguments>
I receive the error
psql: FATAL: role "username" does not exist
where "username" is my OS username (i.e., $USER). It seems like if I were attempting to use pg_dump as myself, it would matter that there's no "username" role, but since I'm not, it shouldn't. How do make pg_dump do something? Postgres 9.4.8 on Fedora. Thanks in advance!
If you are running pg_dump, why does the error message say psql?
I'm guessing you are doing something like this:
sudo -u postgres pg_dump foo bar $(psql blurb blurgle)
or possibly this:
sudo -u postgres pg_dump foo bar | psql blurb blurgle
So you need to sudo -u postgres for both commands.
In any case, it sounds like you are looking in the wrong place for the problem.
Related
I am able to run psql by doing the following:
Davids-d david$ psql --u postgres
Password for user postgres:
psql (9.4.18)
Type "help" for help.
postgres=#
However, when I run the following command, I get an error:
Davids-iMac:datadocs david$ sudo -u postgres psql -f resources/postgresql/initdb.sql
could not identify current directory: Permission denied
What does this mean, and how would I resolve this? Note that I do have the following var set:
david$ echo $PGDATA
/Users/david/PostgreSQL/data/pg94
The issue is the sudo -u postgres.
Your shell is running as you, but you're running the command as the postgres user. It does not have permission to see the file or even be in the current directory.
We can eliminate psql from the equation by just trying to read the file as the postgres user with sudo -u postgres cat resources/postgresql/initdb.sql. You should get the same error.
There's two things you have to do...
cd to a directory that the postgres user can be in.
Put the file in a place the postgres user can access.
/tmp, for example.
Your command seems wrong, try this:
sudo psql -U postgres -f resources/postgresql/initdb.sql
I am having troubles getting started with psql. I can login using the script below
myusername#ubuntu:~/Desktop/dbscripts$ sudo su - postgres
postgres#ubuntu:~$
But, once here, I cannot figure out how to find my .sql file.
I tried the options supplied by Bolo here:
How to import existing *.sql files in PostgreSQL 8.4?
But they only give
myusername#ubuntu:~/Desktop/dbscripts$ psql -U root -d first -f myscript.sql
psql: FATAL: Peer authentication failed for user "root"
and
myusername#ubuntu:~/Desktop/dbscripts$ psql -f myscript.sqlp
sql: FATAL: role "myusername" does not exist
and
myusername#ubuntu:~/Desktop/dbscripts$ sudo su - postgres
postgres#ubuntu:~$ \i myscript.sql
The program 'i' is currently not installed. To run 'i' please ask your administrator to install the package 'iprint'
This can be handled from any of the above options you tried. In which path do you have myscript.sql? After you do su - postgres, give the full path . So it will be psql -d first -f <pathtosqlfile>/myscript.sql. psql -U root will not work unless you have a user root in the database. Try psql -U postgres. You can do \i sqlscript at psql prompt, not at linux command prompt as you have done. The error you are getting "role "myusername" does not exist" can be avoided either by using -U postgres (or any other db user) or by setting the PGUSER environment variable.
I am setting up a website and am having some trouble restoring a database .dump file. I am using centos7, selinux, postgresql 9.4, and apache2.
This is my pg_hba.conf file.
This is the command I am trying to move the dump:
psql --single-transaction -U postgres db_name < dump_location
When I do this, I get the error:
Permission denied.
Am I missing something or is there someway I should alter my settings? Let me know if you need more information.
Thank you!
The operating system user you are running your shell as does not have permission to read the path dump_location.
Note that this is not necessarily the operating system user you run psql as. In a statement like:
sudo -u postgres psql mydb < /some/path
then /some/path is read as the current user, before sudo, not as user postgres, because it's the shell that performs the input redirection, not psql.
If, in the above example, you wanted to read the file as user postgres you would:
sudo -u postgres psql -f /some/path mydb
That instructs psql to open and read /some/path when it's started.
Just make sure that you are using correct database user and you have at least read permission on the dump file.
"psql -d -U postgres -f "
will work.
I'm trying to build a setup script to automate the development environments creation, but I'm having trouble both trying to pipe or using the -c modifier for psql.
I've tried:
sudo su postgres psql -c "ALTER ROLE postgres WITH password 'pass'"
and
sudo su postgres psql -c "ALTER ROLE postgres WITH password 'pass';"
Both of which say "ALTER: command not found"
I've also tried pipe, but I'm not able to combine it with su correctly
eg: I tried something like
sudo su postgres echo "ALTER ROLE postgres WITH password 'pass'" | psql
But postgres can't execute "echo"
And:
echo "ALTER ROLE postgres WITH password 'pass'" | sudo su psql
Which just doesn't work.
So, my first question is: how can I execute this simple command from a sh file?
And the second one, less related: how can I use different users in the commands chained with pipe?
What's wrong is the lack of -c or --command for su to indicate that the rest of the line is a command.
But su is not needed anyway, because there's already sudo. Do this instead:
sudo -u postgres psql -c "ALTER ROLE postgres WITH password 'pass'"
If you are just like me, make sure you are not running this from:
bash-4.2$
but:
postgres=#
I try dropdb mydbname in shell. It do not give any error. But still when I call \l it is still there.
I logged into the postgres server using sudo -u postgres psql.
Other than my main concern I need to know how to go into the database other than just staying outside of it. (as a example if I want to list the tables)
In order to drop database you can use SQL command (but I do not understand why dropdb didn't work) DROP DATABASE mydbname:
sudo -u postgres psql -c "DROP DATABASE mydbname"
It would be good to check if database is not used:
select * from pg_stat_activity where datname = 'mydbname';
The sudo -u postgres psql connects to postgres database. You need to specify database: sudo -u postgres psql mydbname and then you can use metdata commands like \d, \dt, \dv, ...
When you say "shell" ... do you mean the psql shell, not the unix command line shell?
I'd say you're running into this issue:
Postgresql not creating db with “createdb” as superuser, yet not outputting errors
ie you're trying to use dropdb as a psql command, when it's a unix shell command. You're not getting an error because of this:
In psql, why do some commands have no effect?
You didn't terminate the command with a semicolon.
Are you missing the comma(;)? This command worked for me:
drop database <database_name>;
Server should be running, then:
dropdb <database name>
If server is not running, first try:
pg_ctl start -D <mylocal_db_path> -l <mylogfile.log>