This repository provides information on the summit demo3 from 2019 that can be watched here

It contains steps to deploy metalkube and a number of demo workloads.

One node is used as helper node to run the deployment.

Workload Description

• authentication using a dedicated admin user.
• dotnet project with a mssqlserver pod.
• kafka. relies on ceph beeing deployed previously.
• nfs to be used for the windows vm. By default, it requires access to internal network to gather the windows vm although an alternative image can be indicated
• a windows vm connected through the brext bridges of the nodes.
• knative deployment.
• tweeting service using knative.
• uhc modifications so that the cluster shows as baremetal and with a given location at cloud.redhat.com.

Base requirements

You will need to gather:

• ipmi information for your nodes and put in a bm.json file. Use the helper script ipmi.py to validate correct syntax and that credentials are valid.
• a valid pull secret. To reproduce summit demo versions, it will need to be valid for at least for registry.svc.ci.openshift.org, and optionally containing additional credentials for cloud.openshift.com (needed for uhc).
• generate a full config_root.sh.
• decide a CLUSTER_NAME name.
• decide a BASE_DOMAIN name.

You also need your baremetal nodes to have two interfaces, one dedicated for provisioning and the other one for public access.

DNS requirements

for summit demo, we used a dedicated vm running on the helper node with dnsmasq server but in general, one needs the following DNS entries to be resolvable:

• api.$CLUSTER_NAME.$BASE_DOMAIN
• ns1.$CLUSTER_NAME.$BASE_DOMAIN

Additionally, if the nodes are not named master-0, master-1,..., the script fix_ep.sh will need to be called in the middle of the deployment process

Metalkube Master Deployment

the following commands can be used to reproduce the same exact environment as the one used for demo.

Alternatively, follow upstream documentation to deploy kni using a more updated version.

The remainder of this section assumes you want to reproduce summit exact versions

yum -y install git screen bridge-utils net-tools
echo export KUBECONFIG=$HOME/dev-scripts/ocp/auth/kubeconfig >> ~/.bashrc
echo export GOPATH=$HOME/go >> ~/.bashrc
echo export PATH=\$GOPATH/bin:\$PATH >> ~/.bashrc
echo export OS_TOKEN=fake-token >> ~/.bashrc
echo export OS_URL=http://localhost:6385 >> ~/.bashrc
git clone https://github.com/openshift-metalkube/dev-scripts
git checkout 
echo """PATH=/usr/local/bin:/bin
*/5 * * * * $HOME/dev-scripts/fix_certs.sh""" > /tmp/mycron
crontab /tmp/mycron
cd dev-scripts
git checkout a0ea02d9a4e42d0e1e90ec0de7994b558465b4ea
# brext bridge
curl -L https://github.com/openshift-metalkube/dev-scripts/pull/282.patch | git am
# haproxy fix
curl -L https://github.com/openshift-metalkube/dev-scripts/pull/465.patch | git am
# knative fix
curl -L https://github.com/openshift-metalkube/dev-scripts/pull/505.patch | git am

You need to create a config_root.sh file representing your infrastructure.

Use the following commands for an initial sample one:

cp config_example.sh config_root.sh
export PULL_SECRET='{"auths": {"cloud.openshift.com": {"auth": "XXX","email": "XXX"},"quay.io": {"auth": "XXX","email": "XXX"},"registry.svc.ci.openshift.org": {"auth": "XXX"}}}'
sed -i "s/PULL_SECRET=.*/PULL_SECRET='$PULL_SECRET'/" config_root.sh

At this point, you are ready to launch deployment.

make requirements configure repo_sync ironic
eval "$(go env)"
cd $GOPATH/src/github.com/openshift-metalkube/kni-installer
git checkout 1495d7ad68022bd365d98bf6ac846d9f132458bd 
# extend timeout to 60minutes
curl -L https://github.com/openshift-metalkube/kni-installer/pull/52.patch | git am
cd -
make build ocp_run deploy_bmo register_hosts
bash 09_deploy_kubevirt.sh
bash 10_deploy_rook.sh

Metalkube Workers Deployment

Once masters are deployed, additional workers can be added simply by instantiating new baremetalhost custom resources:

oc create -f summit-worker00-cr.yaml -n openshift-machine-api

You can monitor progress by using the following commands:

openstack baremetal node list
oc get baremetalhost -n openshift-machine-api

Additional workloads deployment

For additional workloads deployment, you can either use the scripts provided in this directory or the ones from the extras directory in the dev-script tagged repo:

sed -i 's@bmozaffa/ktweeter:1.0@karmab/ktweeter:v0.3@' 06_tweet.sh
make

Variables

All the variables used for the specific workloads have default values, except for the last four ones, related to tweet service.

Those needs a twitter app to be created

• ADMIN_USER
• ADMIN_PASSWORD
• MSSQL_PASSWORD
• KAFKA_NAMESPACE
• KAFKA_CLUSTERNAME
• KAFKA_PVC_SIZE
• KAFKA_PRODUCER_TIMER
• KAFKA_PRODUCER_TOPIC
• WINDOWS_IMAGE
• VM_SIZE
• VM_MEMORY
• UHC_TOKEN
• CONSUMER_KEY
• CONSUMER_SECRET
• ACCESS_TOKEN
• ACCESS_TOKEN_SECRET