9 Dec 2016

Custom pg_dumpall now works with AWS Redshift

While trying to work with AWS Redshift, it was interesting to see pg_dumpall failing to dump databases in my cluster! 

Delving further, for obvious reasons a managed service like this hides some global-information, which pg_dump(all) needs ... and no, this post is NOT about circumventing that.

This branch, gives a 'near' workaround for those who are okay with being able to extract all databases + (almost*) all Global information (Users etc.) in a single command. 


There are caveats though:

  • Redshift doesn't support COPY TO, so the best workaround is using INSERT. Painful, but works.
  • Barring passwords, all Globals can be dumped. The script just resets all users to be password-less, but that's better than having to do 'CREATE USER ... ' commands for your users by hand!

For some people, these caveats are going to be okay, considering that they get a scriptable way of taking a dump of all databases at one go, along with User information. 

If you're interested in a similar hack for AWS Postgres, you're in luck!

Down the line, I'll try to push this to the core, but for now, this works!

Go Play :) !!

No comments:

Installing pg_tle on Ubuntu: A Quick Guide

Compile & Install pg_tle on Postgres PostgreSQL is a powerful database, but sometimes you want to extend its functionality with custom f...