I do want to search for some data inside a database dump but these dumps are using the binary-compressed format (PGDMP
header).
How can I convert these to SQL without restoring them?
I do want to search for some data inside a database dump but these dumps are using the binary-compressed format (PGDMP
header).
How can I convert these to SQL without restoring them?
pg_restore
, when run without a database name, outputs a text dump to stdout; you can send that elsewhere with -f
or with I/O redirection.
pg_restore -f mydatabase.sql mydatabase.dump
The fastest method that I've used was:
pg_restore mybinaryfile.backup > mysqlfile.sql
No special flags, since pg_restore just spits it out to stdout.
Note that if you run multiple clusters, the restore command may not like the default version...
pg_restore: [archiver] unsupported version (1.12) in file header
In that case you have to specify the version, host and port as in:
pg_restore --cluster 9.1/localhost:5433 -f db.sql db.pgsql
(note that the host:port
info is ignored with the -f option.)
The port (5433) can be determined using the pgsql
command as in:
pgsql --port 5433 template1
When pgsql
connects, it writes a comment such as:
psql (9.3.6, server 9.1.13)
This means you are running pgsql 9.3.6 and that port 5433 references server 9.1.13.
If you are not sure which ports are currently used, you may use the netstat
command as in:
sudo netstat -a64np | grep LISTEN | grep postgres
The sudo
is required for the -p
option which prints the process name. That gives you a list of ports (usually TCP and UDP ports).
Finally, on a Debian/Ubuntu system, you can get a list of installed clusters with the dpkg -l
command as in:
dpkg -l '*postgres*'
The list of entries that start with 'ii' (left most column) are currently installed. You, of course, have similar commands for other Unices to help you determine installed versions.