Copy a table from one database to another in Postgres

Question

I am trying to copy an entire table from one database to another in Postgres. Any suggestions?

Thermonuclear

If you're okay with installing DBeaver, it has a really simple way of transferring between two databases you're connected to. Just right click the source table and select Export Data, target a Database table(s) and set the target as the destination database. – Hasdrubal

@Hasdrubal I'm trying to do the same thing in DBeaver but using dynamic sql. Please let me know if you know how to do it. – Sharasharai

Answer 1

Extract the table and pipe it directly to the target database:

pg_dump -t table_to_copy source_db | psql target_db

Note: If the other database already has the table set up, you should use the -a flag to import data only, else you may see weird errors like "Out of memory":

pg_dump -a -t table_to_copy source_db | psql target_db

Ventriloquist

How will this work for remote-db links? E.g., I need to dump from a different location. – Illiquid

@Illiquid havn't tried this, but I see no reason why it wouldn't work. Try adding user and server specifics to the command, like so pg_dump -U remote_user -h remote_server -t table_to_copy source_db | psql target_db – Ventriloquist

You can try this: "pg_dump -U remote_user -h remote_server -t table_to_copy source_db | psql target_db -U remote_user -h remote_server " – Ironworker

note that if the other database already has the table set up, you should use the -a flag for data only. i.e. pg_dump -a -t my_table my_db | psql target_db. While I'm here, If your database is on a server, I find it easier to just dump the database to a file and then scp that file to the database, then send the contents of the file to psql. e.g.pg_dump -a -t my_table my_db > my_file.sql and after putting that on your server --> psql my_other_db < my_file.sql – Condottiere

How do I get this to work for case sensitive "table_to_copy"? Doesn't work for me when I used: create TABLE "table_to_copy" (arguments...) – Overactive

@EamonnKenny to dump a case-sensitive table, do: pg_dump -t '"tableToCopy"' source_db | psql target_db. Note that single AND double quotes surround the table name – Rheims

@Ventriloquist this is not working for me while running the line in the terminal, have they made changes along these years? – Witkowski

A pitfall is when not all extensions in the source database exist in the target database. You will get many warnings as if there is sql injection.. – Footrope

Is it possible to copy all tables with that are within a certain schema to another remote DB to a identical schema? – Anticathode

Is it possible to copy a table to another database but with a different (existing) schema? Is there any switch I could add to the psql target_db part? – Damron

I get a ton of syntax errors, apparently from psql trying to interpret the content in my fields - ? – Etsukoetta

hey does anyone have any idea, how to implement this method if we are trying to if the two tables (source and target tables) exist on same database but different schemas? – Underglaze

Answer 2

You can also use the backup functionality in pgAdmin II. Just follow these steps:

• In pgAdmin, right click the table you want to move, select "Backup"
• Pick the directory for the output file and set Format to "plain"
• Click the "Dump Options #1" tab, check "Only data" or "only Schema" (depending on what you are doing)
• Under the Queries section, click "Use Column Inserts" and "User Insert Commands".
• Click the "Backup" button. This outputs to a .backup file
• Open this new file using notepad. You will see the insert scripts needed for the table/data. Copy and paste these into the new database sql page in pgAdmin. Run as pgScript - Query->Execute as pgScript F6

Works well and can do multiple tables at a time.

Bellaude

This is a good gui-based solution for moving data between databases. Thanks! – Tandratandy

You can select multiple tables under the Objects section. On OSX, click the SQL button or get the SQL Editor via the Tools menu to paste in the SQL copied from the backup file. – Ja

works, thanks. Very slow though on big tables.. is there a better way to do it to speed it up? (like ignore foreign keys or something?) – Arrogant

@Timothy Here's the postgres documentation page on how to speed up backing up and restoring – Marjie

old answer but still relevant, works great, just don't forget to set Disable triggers when exporting all database – Abercromby

In the background, Pgadmin creates a file holding all necessary SQL INSERT statements with this command: pg_dump --file "/my/path/filename.txt" --host "myhost" --port "myport" --username "myusername" --no-password --verbose --format=p --table "mytable" "mydb" --data-only --column-inserts --inserts – Neuropath

Just remember to check through the other parts of the scripts for anything important like: ALTER TABLE ONLY <table_name> ADD CONSTRAINT <pkey> PRIMARY KEY (field1, field2, field3, etc); You may also see things like : COMMENT ON TABLE <table_name> IS 'This text describes what the table is used for, etc'; Just copy and paste these into the Query Editor along with the CREATE TABLE one. – Rotatory

Answer 3

Using dblink would be more convenient!

truncate table tableA;

insert into tableA
select *
from dblink('hostaddr=xxx.xxx.xxx.xxx dbname=mydb user=postgres',
            'select a,b from tableA')
       as t1(a text,b text);

Canna

Why two dbname in two times..? which one is source and target.? – Febrifugal

tableA that we are inserting to is the destination, and the tableA in the dbLink is the source. – Watchband

if I want to use dblink bun I dont know the structure of the source source table? – Povertystricken

@Povertystricken hey, did you find the answer to your question? – Underglaze

Answer 4

Using psql, on linux host that have connectivity to both servers

( export PGPASSWORD=password1 
  psql -U user1 -h host1 database1 \
  -c "copy (select field1,field2 from table1) to stdout with csv" ) \
| 
( export PGPASSWORD=password2 
  psql -U user2 -h host2 database2 \ 
   -c "copy table2 (field1, field2) from stdin csv" )

Accumulative

No need for export, PGPASSWORD=password1 psql -U ... then you don't even need explicit subshells! Ordinarily, you'll want to do a couple things to set up first, so subshells may be necessary anyway. Also, the passwords won't be exported into subsequent processes. Thanks! – Sternum

@LimitedAtonement Actually you right, export and subshells isn't necessary. It's just a part of more complicated script, and even i didn't try without export and subshells, so, i provide it as is only to be honest and provide worked solution – Accumulative

The table must exist in destination DB. To create it, try pg_dump -t '<table_name>' --schema-only – Archibaldo

Put passwords to ~/.pgpass. – Whist

Answer 5

First install dblink

Then, you would do something like:

INSERT INTO t2 select * from 
dblink('host=1.2.3.4
 user=*****
 password=******
 dbname=D1', 'select * t1') tt(
       id int,
  col_1 character varying,
  col_2 character varying,
  col_3 int,
  col_4 varchar 
);

Jorry

This answer is great because it allows one to filter copied rows (add WHERE clause in the dblink 2nd argument). However, one needs to be explicit about column names (Postgres 9.4) with something like: INSERT INTO l_tbl (l_col1, l_col2, l_col3) SELECT * FROM dblink('dbname=r_db hostaddr=r_ip password=r_pass user=r_usr', 'select r_col1, r_col2, r_col3 from r_tbl where r_col1 between ''2015-10-29'' AND ''2015-10-30'' ') AS t1(col1 MACADDR, col2 TIMESTAMP, col3 NUMERIC(7,1)); (l means local, r is remote. Escape single quotes. Provide col types.) – Reo

Answer 6

If you have both remote server then you can follow this:

pg_dump -U Username -h DatabaseEndPoint -a -t TableToCopy SourceDatabase | psql -h DatabaseEndPoint -p portNumber -U Username -W TargetDatabase

It will copy the mentioned table of source Database into same named table of target database, if you already have existing schema.

Synchronism

Answer 7

Use pg_dump to dump table data, and then restore it with psql.

Mcginnis

Then use another databaserole to connect, a role that has enough permissions. postgresql.org/docs/8.4/static/app-pgdump.html – Manxman

What am I doing wrong? pg_dump -t "tablename" dbName --role "postgres" > db.sql "postgres" would be the user I'm trying to set the role to. It still gives me "Access is denied". – Thermonuclear

Do you have permissions to write the db.sql file? – Rolon

How do I check what permissions I have? – Thermonuclear

This thread is old, but for anyone else having the problem, try using the 'Tools --> Backup' menu in PgAdminIII, which seems to get around the permission problems. – Srinagar

not really a helpful answer, given that the other answers show you how to use pg_dump – Pickax

Answer 8

Here is what worked for me. First dump to a file:

pg_dump -h localhost -U myuser -C -t my_table -d first_db>/tmp/table_dump

then load the dumped file:

psql -U myuser -d second_db</tmp/table_dump

Endearment

for dump load also need "-h localhost" – Newby

I got an error «ERROR: syntax error at or near "ÿ_" --LINE 1: ÿ_-» during the load. Solution was to not use powershell but cmd – Massy

Answer 9

You could do the following:

pg_dump -h <host ip address> -U <host db user name> -t <host table> > <host database> | psql -h localhost -d <local database> -U <local db user>

---

Deterrent

would you like to say something about it – Mae

thats legit 😂 you own me – Mae

Answer 10

To move a table from database A to database B at your local setup, use the following command:

pg_dump -h localhost -U owner-name -p 5432 -C -t table-name database1 | psql -U owner-name -h localhost -p 5432 database2

Displeasure

I tried it. This does not work because you can only give it the first password. – Endearment

@Endearment you can do export PGPASSWORD=<passw> before running the command – Cathrin

Answer 11

Same as answers by user5542464 and Piyush S. Wanare but split in two steps:

pg_dump -U Username -h DatabaseEndPoint -a -t TableToCopy SourceDatabase > dump
cat dump | psql -h DatabaseEndPoint -p portNumber -U Username -W TargetDatabase

otherwise the pipe asks the two passwords in the same time.

Parachronism

Is there possibility that I can mention target database's table name? – Synchronism

Answer 12

I was using DataGrip (By Intellij Idea). and it was very easy copying data from one table (in a different database to another).

First, make sure you are connected with both DataSources in Data Grip.

Select Source Table and press F5 or (Right-click -> Select Copy Table to.)

This will show you a list of all tables (you can also search using a table name in the popup window). Just select your target and press OK.

DataGrip will handle everything else for you.

Inoculum

Please note, DataGrip is a Not Free! – Crain

This functionality is also part of IntelliJ Ultimate (also not free), but something that many people may already have. – Turbine

Answer 13

I tried some of the solutions here and they were really helpful. In my experience best solution is to use psql command line, but sometimes i don't feel like using psql command line. So here is another solution for pgAdminIII

create table table1 as(
 select t1.* 
 from dblink(
   'dbname=dbSource user=user1 password=passwordUser1',
   'select * from table1'  
  ) as t1(
    fieldName1 as bigserial,
    fieldName2 as text,
    fieldName3 as double precision 
  )
 )

The problem with this method is that the name of the fields and their types of the table you want to copy must be written.

Nunci

Answer 14

pg_dump does not work always.

Given that you have the same table ddl in the both dbs you could hack it from stdout and stdin as follows:

 # grab the list of cols straight from bash

 psql -d "$src_db" -t -c \
 "SELECT column_name 
 FROM information_schema.columns 
 WHERE 1=1 
 AND table_name='"$table_to_copy"'"
 # ^^^ filter autogenerated cols if needed     

 psql -d "$src_db" -c  \
 "copy ( SELECT col_1 , col2 FROM table_to_copy) TO STDOUT" |\
 psql -d "$tgt_db" -c "\copy table_to_copy (col_1 , col2) FROM STDIN"

Geof

Answer 15

Check this python script

python db_copy_table.py "host=192.168.1.1 port=5432 user=admin password=admin dbname=mydb" "host=localhost port=5432 user=admin password=admin dbname=mydb" alarmrules -w "WHERE id=19" -v
Source number of rows = 2
INSERT INTO alarmrules (id,login,notifybyemail,notifybysms) VALUES (19,'mister1',true,false);
INSERT INTO alarmrules (id,login,notifybyemail,notifybysms) VALUES (19,'mister2',true,false);

Transcendental

Answer 16

As an alternative, you could also expose your remote tables as local tables using the foreign data wrapper extension. You can then insert into your tables by selecting from the tables in the remote database. The only downside is that it isn't very fast.

Turbine

Answer 17

If the both DBs(from & to) are password protected, in that scenario terminal won't ask for the password for both the DBs, password prompt will appear only once. So, to fix this, pass the password along with the commands.

PGPASSWORD=<password> pg_dump -h <hostIpAddress> -U <hostDbUserName> -t <hostTable> > <hostDatabase> | PGPASSWORD=<pwd> psql -h <toHostIpAddress> -d <toDatabase> -U <toDbUser>

Vernonvernor

Answer 18

for DBeaver tool users, you can "Export data" to table in another database.

Only error I kept facing was because of wrong postgres driver.

SQL Error [34000]: ERROR: portal "c_2" does not exist
    ERROR: Invalid protocol sequence 'P' while in PortalSuspended state.

Here is a official wiki on how to export data: https://github.com/dbeaver/dbeaver/wiki/Data-transfer

Heuer

can upsert work here? – Yl

Answer 19

You have to use DbLink to copy one table data into another table at different database. You have to install and configure DbLink extension to execute cross database query.

I have already created detailed post on this topic. Please visit this link

Benedikta

Answer 20

You can do in Two simple steps:

# dump the database in custom-format archive
pg_dump -Fc mydb > db.dump

# restore the database
pg_restore -d newdb db.dump

In case of Remote Databases:

# dump the database in custom-format archive
pg_dump -U mydb_user -h mydb_host -t table_name -Fc mydb > db.dump

# restore the database
pg_restore -U newdb_user -h newdb_host -d newdb db.dump

Aspectual

Answer 21

Combining this answer and this answer, which is more convenient as you don't need to specify the columns:

TRUNCATE TABLE tableA;

INSERT INTO tableA
SELECT (rec).*
FROM dblink('hostaddr=xxx.xxx.xxx.xxx dbname=mydb user=postgres',
            'SELECT myalias FROM tableA myalias')
       AS t1(rec tableA);

Respirator

Answer 22

It could be done fairly simple manner. Just use the following command

pg_dump –U <user_name> –t <table_name> <source_database> | psql –U <user_name> <targeted_database>

replace values in <> with your specific parameters and also remove <>.

Echinus

Answer 23

On my mac using a | asked for two passwords at the same time which didn't work. here is what I did.

pg_dump -h {host} -U {user} -t {table} {db} | psql postgresql://{user}:{password}@{host}:{port}/{db}

Schnorrer

Answer 24

Just use CREATE TABLE:

CREATE TABLE new_table AS TABLE existing_table;

Jarid

how does this command span two different databases, as posed in the question? – Morven

Answer 25

if you want to copy data from one server database to another server database then you have create dblink connection both database otherwise you can export the table data in csv and import the data in other database table, table fields should be same as primary table.

Crater

Answer 26

Without any piping, on Windows, you can use:

Dump - Edit this to be on one line

"C:\Program Files\PostgreSQL\14\bin\pg_dump.exe"
--host="host-postgres01"
--port="1234"
--username="user01"
-t "schema01.table01"
--format=c
-f "C:\Users\user\Downloads\table01_format_c.sql"
"DB-01"

Restore - Edit this to be on one line

"C:\Program Files\PostgreSQL\14\bin\pg_restore.exe"
--host="host-postgres02"
--port="5678"
--username="user02"
-1
--dbname="DB-02"
"C:\Users\user\Downloads\table01_format_c.sql"

You will be prompted for user passwords.

This solution will put the new table in a schema with the same name (schema01).

Czechoslovak

what is the -1 for? – Galleywest

@Galleywest Single transaction postgresql.org/docs/current/app-pgrestore.html – Czechoslovak

Answer 27

for postgres version >= 8.4.0 the below worked for me

pg_dump -U user -h host --column-inserts --data-only --table=table_name database_name | psql -h host -p port -U user -W database_name

Scoria

Thank you for your interest in contributing to the Stack Overflow community. This question already has quite a few answers—including one that has been extensively validated by the community. Are you certain your approach hasn’t been given previously? If so, it would be useful to explain how your approach is different, under what circumstances your approach might be preferred, and/or why you think the previous answers aren’t sufficient. Can you kindly edit your answer to offer an explanation? – Enumeration

Answer 28

Having done this wrong several times, I'll contribute a solution to SAFELY and RELIABLY copy a table from one remote db to another. There's a lot that can go wrong between the dump and restore. For clarity, some additional criteria in this solution:

• Copy only one table
• Does not delete anything in either source/dest database
• Makes sure the id sequence resumes in the to_table, instead of resetting to 1
• Avoids drop table or --clean mistakes from hasty copy-paste
• Separates dump and restore into two different steps
• Allows flexibility in customizing the to_table (different indexes, etc)
• Both databases are remote
• Each database has a different hostname, port, username, pass

Prerequsites: get pg_dump, pg_restore, psql matching the remote db version

# Figure out which database version is running
#   to use the pg_dump, pg_restore with the version.
# Run the query:
#   select version() # PostgreSQL 14.10

# Then install the matching version
brew tap homebrew/versions
brew search postgresql@
brew install postgresql@14

# Later we can switch back
brew install postgresql@16

Export a table from the remote db, including all large objects in the table

# Dump from 10.0.1.123:1234
#
# -Fc Uses "format custom" optimized for pg_restore
# -b include all large objects, i.e. blobs, bytea, etc
# -U username
# -h hostname
# -p port
# -a only include table data and large objects
# -t table name
# PGPASSWORD is the supported env var to pass in a password
PGPASSWORD="FROM-DB-PASSWORD" pg_dump -Fc -b -U FROM-DB-USERNAME -h 10.0.1.123 -p 1234 -a -t from_table from_db_name > from_table.dump

# Get the last id sequence for restore later
psql -h 10.0.1.123 -p 1234 -d from_db_name -U FROM-DB-USERNAME -W -c "select * from from_table_name_id_seq;"
# last_value == 9999

Import the table into another remote db

# NO CLEAN, NO DROP/DELETE
#
# Safely create a table with a different name for now.
# This helps avoid copy-paste errors accidentally
#   importing back to or deleting things in from_db.
psql -h 10.0.1.456 -p 4567 -d to_db_name -U TO-DB-USERNAME -W -c "create table to_table (id bigserial not null primary key, . . . );"

# Restore to 10.0.1.456:4567
#
# -U username
# -h hostname
# -p port
# -a only include table data and large objects
# -t table name
# -d database name
PGPASSWORD="TO-DB_PASSWORD" pg_restore -h 10.0.1.456 -p 4567 -d to_db_name -U TO-DB-USERNAME -a -t to_table_name from_table.dump

# Restore the id sequence we got from the last export step above.
psql -h 10.0.1.456 -p 4567 -d to_db_name -U TO-DB-USERNAME -W -c "alter sequence to_table_name_id_seq restart with 9999;"

# Rename the table to match the from_table_name
psql -h 10.0.1.456 -p 4567 -d to_db_name -U TO-DB-USERNAME -W -c "alter table to_table_name rename to name_matching_from_table_name;"

# Cleanup
rm from_table.dump

Rely

Answer 29

If you run pgAdmin (Backup: pg_dump, Restore: pg_restore) from Windows it will try to output the file by default to c:\Windows\System32 and that's why you will get Permission/Access denied error and not because the user postgres is not elevated enough. Run pgAdmin as Administrator or just choose a location for the output other than system folders of Windows.

Arroba