Also remove ansible stacks slave as that is out of date.

Per discussion with @sabre1041 @etsauer

Some initial thoughts on how we can add some level of CI to this repository.

1. Assume all quickstarts include an applier inventory (#81 )
2. Set up a travis build to kick off a test of each quickstart in the CASL clusters

• store a service account secret in an encrypted key in travis
• use oc (binary in repo?) to deploy a pipeline to openshift
• utilize 'make-applier-projects-unique' role to prefix all namespaces

@sabre1041 looks like the skopeo utility has made it's way out to at least the fedora channels.

[esauer@localhost ~]$ dnf info skopeo
Last metadata expiration check: 1 day, 11:43:42 ago on Mon Apr 10 21:34:48 2017.
Available Packages
Name        : skopeo
Arch        : x86_64
Epoch       : 0
Version     : 0.1.17
Release     : 1.dev.git2b3af4a.fc25
Size        : 2.1 M
Repo        : updates
Summary     : Inspect Docker images and repositories on registries
URL         : https://github.com/projectatomic/skopeo
License     : ASL 2.0
Description : Command line utility to inspect images and repositories directly on Docker
            : registries without the need to pull them

Worth updating the image?

Creation of a container and deployment artifacts to deploy a syslog server

Phabricator competes in the same space as Github, Gitlab, Gogs, Jira/Confluence/HipChat/Trello/Bitbucket, etc. Phabricator is in use by organizations like Mozilla, Wikimedia Foundation, Facebook, etc. It was originally developed by Facebook [0]. It includes functionality to support Code Reviews, Workboards (Kanban boards), Task Management, Wikis, and Chat [1].

[0] https://en.wikipedia.org/wiki/Phabricator
[1] https://phacility.com/phabricator/

Create a containerized Zabbix server

openshift/jenkins#688

The generic template defaults to using Dockerfile.rhel7, while the jenkins-slave-ansible doesn't have this file and hence requires the DOCKERFILE_PATH to be set as part of the params.

Probably should align jenkins-slave-ansible with the other slave and have a Dockerfile.rhel7 file.

Need to allow for the /etc/passwd file to be updated on startup so that the randomly chosen UID from OpenShift is integrated into /etc/passwd.

We would prefer to have separate Dockerfiles as the desire is to put these under autobuilds so that users don't have to build the images to consume them. We already do a decent job of auto publishing the CentOS versions of the images to docker cloud, and we don't want to have to make a repo change to set up a build in OpenShift. If it becomes that we are repeating 90% of the code between the two, let's look to use common RUN scripts to reduce that.

To align with reccommendations from the upstream Kubernetes community, convert the existing JSON based configurations (templates, misc API objects) to YAML format

Have a client that has requested a builder image that would include rhel7+/openjdk8+/gradle4+. They use gradle for their builds and are looking to transition a bunch of new projects to OpenShift, but need this new image.

Parts of this live in labs-ci-cd today: https://github.com/rht-labs/labs-ci-cd/tree/master/openshift-templates/gogs

Right now we are using 'latest' tag for every single Jenkins Slave container image. We should use the corresponding existing tag aligned with the OCP version and the 'cut release' we use for this repository. We have recently identified some issues while using the 'latest' tag

While following steps to configure Kafka on Openshift, facing an error "Failed to pull image "172.30.1.1:5000/kafka/kafka:latest": rpc error: code = 2 desc = Error: image kafka/kafka:latest not found".
Zookeeper deployment is also failing with error as "Failed to pull image "172.30.1.1:5000/kafka/zookeeper:3.4.11": rpc error: code = 2 desc = Error: image kafka/zookeeper:3.4.11 not found"

Images using RHEL 7.4 do not properly implement yum-config-manager as seen in this BZ

https://bugzilla.redhat.com/show_bug.cgi?id=1479388

Many of our images use yum-config-manager and will not be able to properly build images.

An alternate solution would be to enable repositories during the yum install command

cc: @sabre1041

Overview

Add the ability for Jenkins to run a slave which can build Erlang applications using Rebar3. Also, add an S2I image build which can convert an Erlang release to a runnable container image using Rebar3

Acceptance criteria

• A new Jenkins slave image which has Rebar 3 and Erlang/OTP installed for use with creating a "release"
• A new S2I image which can take an Erlang release created via Rebar 3 and create a runnable container

Some base images have repos enabled (or not enabled) that doesn't align with the host running the container. This should be investigated to figure out how we can avoid using --enablerepo or --disablerepo within the Dockerfile.

More details here:
#98 (comment)

Hi,

Is there any Skopeo template available for OpenShift Origin cluster?

I want to deploy a Skopeo component on Origin cluster and have a capability of moving images from OpenShift Origin cluster's docker registry to OpenShift Enterprise cluster's docker registry.

Thank you in advance.

Some desired features:

• Make sure we can still auto-update when base images change.
• Build/Deploy a spring boot app
• AB services
• Enable autoscaling
• Basic integration/unit testing
• Better triggering with Git (probably through git server-specific jenkins build plugins - https://wiki.jenkins-ci.org/display/JENKINS/GitHub+pull+request+builder+plugin)
• Multi-feature branch support
• Multi-cluster promotion using skopeo slave image (https://github.com/redhat-cop/containers-quickstarts/tree/master/jenkins-slaves/jenkins-slave-image-mgmt)
• Provision multiple pipelines from a "master jenkins", with ansible-stacks (https://github.com/redhat-cop/containers-quickstarts/tree/master/jenkins-slaves/jenkins-slave-ansible-stacks)
• Idempotency in the pipeline scripe

@redhat-cop/containerize-it

The current 3.X releases of the applier are deprecated and will be removed sometime in the future. A number of quickstarts are still referencing these deprecated tags in the requirements.yml files. It would probably be best to get these updated to an appropriate version of the applier moving forward (either 1.x or 2.x)

So modules need to use pypi libs, so this is a good feature to ship since we already have epel installed.

Current use case is rht-labs/labs-ci-cd#212, which is a worked around for redhat-cop/openshift-applier#47

OpenShift 3.6 introduced mechanisms to reference ImageStreams from StatefulSets.

https://docs.openshift.com/container-platform/3.6/release_notes/ocp_3_6_release_notes.html#ocp-36-container-orchestration

Update existing implementations to make use of this functionality

The GitLab template doesn't support supplying custom certificates as-is, and it should.

Parts of this live in labs-ci-cd today: https://github.com/rht-labs/labs-ci-cd/tree/master/openshift-templates/gitlab

Support for utilizing the openshift-applier to provision applications

• Creation of inventory for quickstarts
• Documentation on usage
• Continuation of support without openshift-applier

As a DevOps team member, I need to be able to run Selenium tests from within Jenkins pipelines. In order to run Selenium tests, I need a Jenkins Agent which can execute Google Chrome and Firefox web browsers in a "headless" environment using Xvfb running as a non-root user.

oc process -f ../templates/jenkins-slave-image-mgmt-template.json | oc apply -f-

Should be

oc process -f ../templates/jenkins-slave-image-mgmt-template.yml | oc apply -f-

in
/container-quickstarts/jenkins-slaves/jenkins-slave-image-mgmt/README.md

We need to define what a "container quickstart" is. I see this looking similar to https://github.com/redhat-cop/container-pipelines/blob/master/README.md

Some ideas for requirements:

• CentOS (or other community base) for hosting on public registry (currently dockerhub)
• RHEL (or Atomic) base for enterprise consumption
• Applier inventory that proves functionality of the quickstart
• README that includes:
  • Requirements & Assumptions section
  • Quickstart Instructions for getting it up and running
  • Configuration section that talks about any customization options
  • Verification instructions that show the quickstart is working as expected
  • Tear Down commands

When running on RHEL hosts, the jenkins-slave-ansible often uses a rather old 2.4.x version of Ansible. This is problematic for various tools, including the openshift-applier, which relies on ansible version >=2.5.x.

Current Ansible version is 2.6, so this issue is to get the jenkins-slave-ansible updated to use Ansible version 2.5 or greater.

@sabre1041
please update from petset to statefulset
remove the annotation

Looking for a ready to go Hygieia OCP template...

Dear all,

I followed the instruction reported on the readme, but it seems that during the deployment of zookeeper-0, the pod fails to mount the volume. Into the event log i got this error:

  Unable to mount volumes for pod "zookeeper-0_zookeeper(2a0c3a72-3727-11e8-8e85-005056a31518)": timeout expired waiting for volumes to attach/mount for pod "zookeeper"/"zookeeper-0". list of unattached/unmounted volumes=[datadir]

MountVolume.SetUp failed for volume "pv0004" : mount failed: exit status 32 Mounting command: systemd-run Mounting arguments: --description=Kubernetes transient mount for /var/lib/origin/openshift.local.volumes/pods/2a0c3a72-3727-11e8-8e85-005056a31518/volumes/kubernetes.io~nfs/pv0004 --scope -- mount -t nfs lv170170.vtlab.local:/nfsshare/pv0004 /var/lib/origin/openshift.local.volumes/pods/2a0c3a72-3727-11e8-8e85-005056a31518/volumes/kubernetes.io~nfs/pv0004 Output: Running scope as unit run-22561.scope. mount.nfs: No route to host

Do you have idea how I can solve the problem? Thanks in advance.

Probably makes sense to start here: https://github.com/sclorg/s2i-nodejs-container

Any plans for adding an eclipse che container?

The following entry is pointing to a Dockerfile which doesn't exists

https://github.com/redhat-cop/containers-quickstarts/blob/master/jenkins-slaves/templates/jenkins-slave-generic-template.yml#L81

It should be Dockerfile and not Dockerfile.rhel7

In the s2i-liberty image it would be great to make the copy of the config directory recursive which would allow things like including database drivers, etc where you want a more structured file system rather than just dump everything in config (which ends up in usr/servers/defaultServer). To make this change just do the following change in s2i/bin/assemble-runtime:

cp $S2I_DESTINATION/src/config/* /config

To include the -r switch:

cp -r $S2I_DESTINATION/src/config/* /config

@sabre1041 please update kafka to statefulset

As per the title; in labs we have the owasp-zap jenkins slave in a separate repo. Wasn't caught as part of #78. Are we cool to lift and shift it to here with the other jenkins-slave-*?

https://github.com/rht-labs/owasp-zap-openshift

I'm happy to do the leg work on it!

Based on @eformat's work https://github.com/eformat/openshift-apicurio

I'm working on a PR now

If I follow the s2i-jekyll readme, I get successful builds for both the base image and the example openshift-playbooks app, however, when the application container starts, I see a crashloop with the following log output:

`/` is not writable.
Bundler will use `/tmp/bundler/home/unknown' as your home directory temporarily.
bundler: command not found: jekyll
Install missing gem executables with `bundle install`
--

Since #87 was fixed for just one image jenkins-slave-golang the jenkins-slave-generic-template.yml has been defaulted to look for a Dockerfile.rhel7. This means instructions for over ridding this need to be included on the README of each slave that does not have a Dockerfile.rhel7 or a Dockerfile.rhel7 version needs to be included with each slave.

Also any connecting repositories using the jenkins-slave-generic-template.yml template via the openshift-applier should also be updated / version pinned accordingly.

@sabre1041 ragarding zookeeper
requiring to run with user 1001 creates issues at some customers.
if this is necessary can the reason be explained in the documentation.
Also can we explain why the fsgroup is necessary?

We should be able to replace a good amount of shell logic with standard pipeline DSL methods from the OpenShift Pipelines Plugin.

https://github.com/jenkinsci/openshift-pipeline-plugin

Strimzi provides a more mature method of deploying Kafka on OpenShift. As a result, the quickstart should be deprecated