copr-setup.txt

WORK IN PROGRESS

- How coprs works
  - SEE README FOR OVERVIEW


- need at least 1 server
- some instances/systems for builders (Fedora Copr use OpenStack)

Backend:
ansible playbook: 
  https://infrastructure.fedoraproject.org/infra/ansible/playbooks/groups/copr-backend.yml

- enable epel (if you use RHEL)
- enable optional channel (if you use RHEL)
- install copr-backend package
- make sure that Redis is up and running
- edit /etc/copr/copr-be.conf
  - change results_baseurl and frontend_url to your hostname
  - frontend_auth must have same value as BACKEND_PASSWORD from have frontend in /etc/copr/copr.conf
  - define your build groups: (more info in the config comments)
    - spawn_playbook - path to ansible playbook which spawns builder
    - terminate_playbook - path to ansible playbook which terminate builder
      - you can see /usr/share/doc/copr*/playbooks/ for examples
      - for Fedora copr playbook see FEDORA PLAYBOOK below
  - other variables can be edited as well, but it is optional
- ntp should be enabled and started
- ports 22, 80, 443 should be open
- cp /usr/share/doc/copr-*/lighttpd/* /etc/lighttpd/conf.d/
- edit file /etc/lighttpd/lighttpd.conf as follows:
       -server.document-root = server_root + "/lighttpd"
       +server.document-root = "/var/lib/copr/public_html"
   this is recommended to be separate mount point

FEDORA PLAYBOOK:
1. spawn instance playbook - 
   https://infrastructure.fedoraproject.org/infra/ansible/roles/copr/backend/files/provision/builderpb_nova.yml
   - just an ansible playbook which sets up and provisions an instance
     in our case it is using eucalyptus but it doesn't have to be
     any cloud provider - it just needs to be able to get an ip an
     ssh into it - cloud providers or private clouds just make more 
     sense
2. provisioning dir
   https://infrastructure.fedoraproject.org/infra/ansible/roles/copr/backend/files/
   - what gets pushed to the other side - ssh keys, users, mock
     configs, etc
3. terminate instance playbook -
   https://infrastructure.fedoraproject.org/infra/ansible/roles/copr/backend/files/provision/terminatepb_nova.yml
4. create ~copr/.ssh/id_rsa with ssh key to VM guests


Frontend:
- if you run on EPEL - add epel repo
- add yum repository http://copr-be.cloud.fedoraproject.org/results/msuchy/copr/
- yum install copr-frontend copr-selinux

- apache setup
  cp /usr/share/doc/copr-frontend/coprs.conf.example /etc/httpd/conf.d/coprs.conf
  and edit it to your needs - at least ServerName, port and WSGIScriptAlias

- if you run on EPEL add virtenv of flask and friends
  - virtualenv --distribute virtenv  
  - source virtenv/bin/activate
  - pip install flask flask-sqlalchemy flask-openid flask-wtf alembic
  - pip install -I  SQLAlchemy MarkupSafe 
  - pip install pytest

- edit credentials /etc/copr/copr.conf (or /etc/copr/copr_devel.conf for development mode)
- cd /usr/share/copr/coprs_frontend
- configure DB first - see DB section below
# for develop setup do:
COPR_CONFIG=/etc/copr/copr_devel.conf ./manage.py create-db --alembic alembic.ini
# for production setup do:
./manage.py create-db  --alembic alembic.ini

- for developing you can run:
  ./manage.py runserver -t <FQDN>
  and this will load DevelopmentConfig from coprs_frontend/coprs/config.py

- for production is used ProductionConfig from coprs_frontend/coprs/config.py

- Add chroots. E.g:
 su - copr-fe -c './manage.py create-chroot fedora-18-x86_64 fedora-18-i386'


DB config
=========

Copr can use either SQLite or PostgreSQL. Choose one.

SQLite
~~~~~~

- yum install sqlite
- in file /etc/copr/copr.conf put:
  SQLALCHEMY_DATABASE_URI = 'sqlite:////var/lib/copr/data/copr.db'
This will use sqlite db file in /var/lib/copr/data/copr.db (manage.py will create that file)
- Note that you have to either create db as copr-fe user or you have to run:
  chown copr-fe:copr-fe /var/lib/copr/data/copr.db
  after you run "./manage.py create-db --alembic alembic.ini"

PostgreSQL
~~~~~~~~~~

- yum install -y postgresql-server
- chkconfig postgresql on
- on Fedoras, run:
 postgresql-setup initdb
- everywhere else run:
 service postgresql initdb

- service postgresql start
- su - postgres -c 'PGPASSWORD=coprpass ; createdb -E UTF8 coprdb ; yes $PGPASSWORD | createuser -P -sDR copr-fe'

Configure the user to use md5 password to connect to that database. Put the lines like following to /var/lib/pgsql/data/pg_hba.conf. Avoid the common pitfall: Make sure you put them *before* those existing lines that are for all..

local coprdb copr-fe md5
host  coprdb copr-fe 127.0.0.1/8 md5
host  coprdb copr-fe ::1/128 md5
local coprdb postgres  ident

Then reload PostgreSQL:
 service postgresql reload

and test the connection:

PGPASSWORD=coprpass psql -a -U copr-fe coprdb
PGPASSWORD=coprpass psql -a -h localhost -U copr-fe coprdb

Then run:
  "./manage.py create-db --alembic alembic.ini"

Upgrade
=======
After upgrade of copr-frontend you should upgrade db schema:
cd /usr/share/copr/coprs_frontend
alembic upgrade head

Make yourself admin
===================

Run:
  cd /usr/share/copr/coprs_frontend/
  ./manage.py alter-user --admin msuchy

Then navigate to:
  http://your.copr.fqdn/admin

Add new chroot
==============

First create new chroots:

    python manage.py create-chroot fedora-24-i386 fedora-24-x86_64

And then copy data from rawhide into them:

    python manage.py rawhide-to-release fedora-rawhide-i386 fedora-24-i386
    # similarly for fedora-24-x86_64

DEBUGGING
=========

Run mockremote as standalone process:
su - copr
cd /usr/share/copr
python backend/mockremote.py -b 172.16.3.4 -r fedora-19-x86_64 <url of src.rpm>