Grant Gainey edited this page Dec 12, 2016 · 2 revisions
Clone this wiki locally

For up-to-date information about the PostgreSQL port, see PostgreSQL. This page is now obsolete.

Postgresql Quick Start Guide

<div style="background-color:#2b299e; color:white; background-image:url('/spacewalk/attachment/wiki/WikiSnippetsAndTemplates/workinprogress.png?format=raw'); background-repeat:no-repeat; padding-left:7em;background-position:1em;" >
   <h2>Work in Progress</h2>
   <div style="padding-bottom:0.5em;" >
       <p>This is page in currently in a draft state.</p>

Want to help fix Spacewalk database queries so that they work in Postgres? Getting started is easy.

I. Getting everything installed

  • Install git, if you don't have it installed: (may require EPEL repositories for CentOS/RHEL 5)

    yum install git }}}

    • Clone spacewalk from git {{{ git clone git:// }}}
    • Install postgres, if you haven't already. You will need 8.4 beta because the schema has not been back-ported to earlier versions. The easiest way to in stall this version is using this public yum repo []. {{{ yum install postgresql-server }}}
    • Install the python lex/yacc parser and PostgreSQL libraries: {{{ yum install python-ply python-pgsql }}}
    • Install chameleon, the tool that builds both Oracle and Postgres schema from a single meta-schema {{{ yum install chameleon }}} Or, {{{ rpm -Uvh }}}
    • Formerly: dogchow
    • Get the right branch, where the postgres porting work is going on {{{ cd spacewalk git checkout origin/pgsql }}}
    • Configure PostgreSQL:
    • As root: {{{ /sbin/service postgresql initdb

    Setup PostgreSQL authentication

    TODO: Probably need some careful thought here in the future:

    perl -p -i -e 's/ident sameuser/trust/g' /var/lib/pgsql/data/pg_hba.conf perl -p -i -e 's/ident/trust/g' /var/lib/pgsql/data/pg_hba.conf /sbin/service postgresql start createuser --user postgres spacewalk --superuser createdb --user spacewalk spacewalk && createlang --user spacewalk plpgsql spacewalk psql --user spacewalk spacewalk # to verify that the spacewalk db has been successfully created }}}

    • Build schema! We hope to improve this and make it simpler soon. {{{ cd spacewalk/schema/spacewalk ../../rel-eng/bin/tito build --rpm --test


    cd schema/spacewalk/postgres make psql -U postgres -f main.sql }}}

    • Look at chameleon go, creating both Oracle and Postgres schema on the fly! Warnings are fine.
    • Also runs blend, which does depsolving and creates a properly ordered set of create scripts.
    • Examine output for the location of the noarch RPM that was built and run rpm -Uvh RPM.
    • Deploy PostgreSQL specific modifications not yet packaged:
    • Because we cannot properly tag and build rpms while working in a git branch, we've setup a short shell script that scp's all modified files into their correct locations. {{{ SWHOST=root@localhost scripts/ el5 }}}
    • Use "f10" instead of el5 if deploying to Fedora 10.
    • NOTE: Using scp, technically not required with these instructions as we're running the script on the host.
    • The script is setup for Fedora 10. If you need to install on EL5 then you'll need to modify some simple perl directories in the script.
    • Setup an answers file for PostgreSQL: {{{ cat > ~/answers-postgresql.txt <<'EOF' admin-email = db-backend = postgresql db-user = spacewalk db-host = localhost db-password = spacewalk db-sid = spacewalk db-port = 5432 db-protocol = TCP ssl-set-org = My Org ssl-set-common-name = My Org Test ssl-set-org-unit = BLAH ssl-set-city = Raleigh ssl-set-state = NC ssl-set-country = US ssl-password = wer tyu ssl-set-email =


  • Install! {{{ spacewalk-setup --disconnected --answer-file=~/answers-postgresql.txt }}}

  • THIS WILL FAIL, but it goes a long way before it does. :) Schema will be populated.

  • If you want to try again, you must destroy your spacewalk database, we cannot "clear" yet. {{{ dropdb --user spacewalk spacewalk && createdb --user spacewalk spacewalk && createlang --user spacewalk plpgsql spacewalk }}}

II. Fixing a query, an example

Your process may be different, but this is my process for porting a query, step-by-step.

  • Go find a file with some queries in it. There's tons. :) At some point, we will have a wiki page that will allow contributors to claim a file. For now, I'm picking a file at random: spacewalk/backend/server/ A good way to find queries in the Python codebase, for instance: grep -r rhnSQL ..
  • Hmm, looks like grep -r "#pgport" . does a better job. Why are these tags invalid, again? Is there any way to rescue any of this work?