Skip to content

Latest commit

 

History

History
249 lines (224 loc) · 7.98 KB

USAGE.md

File metadata and controls

249 lines (224 loc) · 7.98 KB
Raw

Some examples and use-cases

I want to enable masquerade on the undercloud

  1. Create a "masquerade.yaml" in local_env directory
  2. Add the following content:
---
undercloud_config:
  - section: "ctlplane-subnet"
    option: "masquerade"
    value: "true"
  1. Run the playbook with the following options:
ansible-playbook builder.yaml -t lab -e @local_env/masquerade.yaml

I want to test that patch from gerrit

  1. Fetch the wanted patch in your local repository
  2. Follow "I want to test that local change on my computer"

I want to test that local change on my computer

  1. Create a "patch.yaml" in local_env directory
  2. Add the following content:
---
synchronize:
  - name: project-name
    base: /home/USERNAME/working/base/dir
    dest: /home/stack/tripleo/
  1. Run the playbook with the following options:
ansible-playbook builder.yaml -t lab -e @local_env/patch.yaml

I want to install that RPM in addition

  1. Create "rpm.yaml" in local_env directory
  2. Add the following content:
custom_rpms:
  - rpm_name
  - https://some/remote/rpm
  1. Run the playbook with the following options:
ansible-playbook builder.yaml -t lab -e @local_env/rpm.yaml

I want to ensure yum/dnf is using the same mirror

This can be handy if you have a caching proxy, or if you have local mirrors

  1. Create a new env file, for instance "centos-repositories.yaml"
  2. add custom_repositories as follow:
custom_repositories:
  - name: BaseOS
    file: CentOS-Base
    uri: your-unique-mirror
    priority: 100
  - name: AppStream
    file: CentOS-AppStream
    uri: your-unique-mirror
    priority: 100
  - name: extras
    file: CentOS-Extras
    uri: your-unique-mirror
    priority: 100

⚠️ This setting is used on both the nat-vm and undercloud nodes. Therefore you can't use it when deploying centos-7 undercloud, since the nat-vm is running on centos-8.

I want to mimic a slow system

  1. Create your vm listing and add some iotune options
  2. Enjoy a slow system You can take the ./environments/standalone-slow.yaml as an example. A deploy generate a constant load of 3-4 with that setup on my builder, ensuring you will hit any race conditions/timeouts you can find in upstream CI. Although those values are, probably, a bit too harsh :).

I want to customize my tmate

  1. create local_envs/tmate.yaml
  2. push something like that in it:
tmate_config: |
  set -g tmate-server-host "tmate.your-domain.tld"
  set -g tmate-server-port "2200"
  set -g tmate-server-rsa-fingerprint "SHA256:XXXX"
  set -g tmate-server-ed25519-fingerprint "SHA256:YYYY"

I want to push some custom semodule

  1. Create my-module.te file somewhere on your system
  2. Create semodule.yaml in local_env directory
  3. Add the following content:
semodules:
  - name: my-module
    src: /path/to/my-module.te
  1. Run the playbook with the following options:
ansible-playbook builder.yaml -t lab -e @local_env/semodule.yaml

I want to inject that RPM/patch/other in my containers

Tripleo-lab relies on tripleo-image-modify for this part. In order to inject things into the container image prior any deploy, you have to know a bit about therelated doc.

For instance, if you want to build RPM based on a serie of patches and inject them into a bunch of container images:

  1. create local_env/patches.yaml
synchronize:
  - name: mistral
    container: true
    base: /home/cjeanner/work/gerrit
    dest: /home/stack/tripleo/
  1. Create local_env/container-modify.yaml
modify_image:
  - push_destiation: true
    includes:
      - mistral-api
      - mistral-engine
      - mistral-event-engine
      - mistral-executor
    modify_role: tripleo-modify-image
    modify_append_tag: '-dev'
    modify_vars:
      tasks_from: rpm_install.yml
      rpms_path: /home/stack/container-rpms
  1. Run the deploy command, including those two files:
ansible-playbook builder.yaml -e @local_env/container-modify.yaml -e @local_env/patches.yaml

I want to deploy Red Hat Director

Please note this is for Red Hat employees and was NOT tested for any other use

  1. Create a "rhel.yaml" in local_env directory
  2. Add the following content:
---
base_image: rhel
rhos_release_repo_url: http://xxxxxx
proxy_host: xx.yy.zz.aa:PORT
undercloud_sample: /usr/share/instack-undercloud/undercloud.conf.sample
# Set up container things
rhos_containers_registry: <some address>
  1. Run the playbook with the following options:
ansible-playbook builder.yaml -t lab -e @local_env/patch.yaml

Notes:

  • The proxy_host is discussed in this blog post
  • The rhos_release stuff is internal to Red Hat
  • This will not take into account any registration to RHEL
  • The undercloud_sample is mandatory as long as instack-undercloud is used.

I want to use a local registry and a Squid proxy for the packages

This is especially interesting on slow connection, or if you just want to avoid fetching 10G for each deploy. In order to do so, you can run a docker-registry on your builder, as well as a squid proxy.

Registry

The docker-registry can be launched like this:

podman create --name registry \
  --net=host -d \
  -v /srv/slow/registry:/var/lib/registry \
  --conmon-pidfile=/srv/slow/registry/container.pid \
  -e=REGISTRY_HTTP_ADDR=192.168.122.1:5000 registry:2

This will create the podman container, and add persistent storage using a dedicatec volume. Of course, update things accordingly. Beware of SELinux!

Then start the container, using podman start registry and you're set.

Note for CentOS-8 builder: firewalld needs some love:

$ sudo firewall-cmd --zone=libvirt --add-service=docker-registry --permanent

This will actually allow your VMs to access the registry. Please note, if you decide to use another port than the default, you'll need to tweak the firewall-cmd as well.

Squid

Here, you want to ensure things are actually getting cached. You can push this in your squid configuration:

# Leave coredumps in the first cache dir
coredump_dir /var/spool/squid
cache_dir ufs /srv/slow/squid 20000 16 256
maximum_object_size 5 GB
cache_replacement_policy heap LFUDA


refresh_pattern -i .rpm$ 129600 100% 129600 refresh-ims override-expire
refresh_pattern -i .iso$ 129600 100% 129600 refresh-ims override-expire
refresh_pattern -i .deb$ 129600 100% 129600 refresh-ims override-expire

Note the cache_dir_ufs - I'm using a custom volume, you might want to update this path as well.

Note for CentOS-8 builder: firewalld needs some love:

$ sudo firewall-cmd --add-service=squid --zone=libvirt --permanent

This will ensure your VMs have access to this service. Here again, if you put a custom port for squid, you'll need to tweak a bit the firewalld command.

Use them all!

Now you'll need to instruct tripleo-lab to use those two services. For Squid, create some local_env/proxy.yaml:

---
proxy_host: 192.168.122.1:3128
custom_repositories:
  - name: BaseOS
    file: CentOS-Base
    uri: http://miroir.univ-lorraine.fr/centos/$releasever/BaseOS/$basearch/os/
    priority: 100
  - name: AppStream
    file: CentOS-AppStream
    uri: http://miroir.univ-lorraine.fr/centos/$releasever/AppStream/$basearch/os
    priority: 100
  - name: extras
    file: CentOS-Extras
    uri: http://miroir.univ-lorraine.fr/centos/$releasever/extras/$basearch/os
    priority: 100

For the registry, create some local_env/registry.yaml:

---
container_prepare_overrides:
  ceph_namespace: 192.168.122.1:5000/ceph
  ceph_prometheus_namespace: 192.168.122.1:5000/prom
  namespace: 192.168.122.1:5000/tripleomaster

And just pass those two env files to your deploy. Enjoy faster deploy!