Create a .pgpass file if gitlab_rails['db_password'] is not nil
I think this is good.
Still need to update https://gitlab.com/help/administration/high_availability/database.md as it's reference to postgresql['sql_password'] is not accurate
Fixes #1392 (closed)
Merge request reports
Activity
assigned to @ibaum
assigned to @twk3
changed milestone to %8.15
assigned to @ibaum
assigned to @ibaum
added 1 commit
- 064ff0b9 - Fix chef failure when PostgreSQL user doesn't exist yet
assigned to @twk3
assigned to @ibaum
added 1 commit
- 5e733595 - Move to using lazy loading for pgpass template instead of if statement
assigned to @twk3
183 183 action :nothing 184 184 not_if { !pg_helper.is_running? || pg_helper.is_slave? } 185 185 end 186 187 template 'postgresql user pgpass' do 188 path lazy { "#{node['etc']['passwd'][postgresql_user]['dir']}/.pgpass" } 189 source 'pgpass.erb' 190 owner postgresql_user 191 mode '0600' 192 not_if { node['gitlab']['gitlab-rails']['db_password'].nil? } I built this (if you push your branch to dev, it will spit out builds you can test) and played with it a bit. And I don't think pgpass is a fix for the issue the customer was having. I've left a comment in the issue: https://gitlab.com/gitlab-org/omnibus-gitlab/issues/1392#note_20046720 and lets get some feedback from @marin
@marin Seeing as nobody else is asking for it, I think we should hold off on merging for now. It's not my first choice of authentication methods for #1807 (closed) but I can't rule it out quite yet.
I vote we cancel the merge request, but leave the branch in place for now.
Sounds good @ibaum
