commit	a9e53c91e4ca773863bae4c66df9a560280e416c	[log] [tgz]
author	Tinoj Joseph <Tinoj.Joseph@radisys.com>	Tue Aug 23 10:08:46 2022 +0530
committer	Tinoj Joseph <Tinoj.Joseph@radisys.com>	Wed Aug 24 16:08:42 2022 +0530
tree	b1e7072ac4efa83442fff60f86e7a5bf810b09ad
parent	dd1fd9d86e0b375192958a115d34070c04a6f1a6 [diff]

tree: b1e7072ac4efa83442fff60f86e7a5bf810b09ad

README.md

Helm Charts to Deploy VOLTHA 2.x

This repository defines Kubernetes Helm charts that can be used to deploy a VOLTHA instance. More information and documentation can be found in the voltha docs.

Installing charts

To deploy VOLTHA a Kubernetes environment is required. There are many mechanisms to deploy a Kubernetes environment and how to do so is out of scope for this project. A Simple search on the Internet will lead to the many possibilities.

In addition to a base Kubernetes in order to pass traffic to an OLT additional services that are external to VOLTHA are required, such as an OpenFlow Controller with applications to support authentication (EAPOL) and IP address allocation (DHCP) as examplified by the SEBA Project.

Example deployment

The following describes how to deploy VOLTHA.

Prerequisites

To use the charts for VOLTHA the following Helm repositories should be added to your helm environment:

helm repo add onf https://charts.opencord.org
helm repo update

Installing VOLTHA infrastructure

VOLTHA relies to a set of infrastructure components (ONOS, Kafka, ETCD, ...) that can be installed via the voltha-infra helm chart:

helm upgrade --install --create-namespace -n infra voltha-infra onf/voltha-infra

By default the voltha-infra helm chart will install one instance of each component, but that can be customized via a custom value file or via flags, eg:

helm upgrade --install --create-namespace -n infra voltha-infra onf/voltha-infra \
  --set onos-classic.replicas=3,onos-classic.atomix.replicas=3 \
  --set kafka.replicaCount=3,kafka.zookeeper.replicaCount=3 \
  --set etcd.statefulset.replicaCount=3

To deploy a released version of the voltha-infra chart you can specify the --version in the command. As an example to deploy the infrastructure for the 2.8 LTS release you can use --version 2.8.0 like:

helm upgrade --install --create-namespace -n infra --version 2.8.0 voltha-infra onf/voltha-infra

Accessing the ONOS Cli and API

In order to access the ONOS CLI you need to expose the ONOS SSH port:

kubectl -n infra port-forward --address 0.0.0.0 svc/voltha-infra-onos-classic-hs 8101:8101

Once that is done you can ssh into ONOS by:

ssh karaf@127.0.0.1 -p 8101

To access the ONOS Rest API and GUI you need to expose:

kubectl -n infra port-forward --address 0.0.0.0 svc/voltha-infra-onos-classic-hs 8181:8181

Customizing the ONOS configuration

The ONOS configuration is defined in two separate variables in the value file:

onos-classic.config.netcfg: multiline text (json)
onos-classic.config.componentConfig: yaml list of component name and multiline text

Being the content of the configuration multiline text is not possible to override the configuration via --set it's necessary to create a custom value file with your content and install the chart with:

helm upgrade --install --create-namespace -n infra voltha-infra onf/voltha-infra -f myfile.yaml

Support for logging and tracing (optional)

VOLTHA comes with support for Jaeger and EFK (Elastic, Fluentd, Kibana) integration. In order to deploy those components together with the infrastructure add these flags to the command to install the VOLTHA infrastructure:

helm upgrade --install --create-namespace -n infra voltha-infra onf/voltha-infra \
  --set voltha-tracing.enabled=true \
  --set efk.enabled=true

Once kibana is running execute this command to properly configure it:

 curl -v -X POST -H Content-type:application/json -H kbn-xsrf:true http://localhost:5601/api/saved_objects/index-pattern/logst* -d '{"attributes":{"title":"logst*","timeFieldName":"@timestamp"}}'

NOTE In order to send a request to kibana you need to expose the port with kubectl port-forward -n infra --address 0.0.0.0 svc/voltha-infra-kibana 5601

Deploy VOLTHA

VOLTHA encompass multiple components that work together to manage OLT devices. Such group of component is known as a stack and is composed by:

VOLTHA core
OfAgent (OpenFlow Agent)
OLT Adapter
ONU Adapter

To deploy a VOLTHA stack with the opensource adapters (OpenOLT and OpenONU) you can use the voltha-stack chart:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra

We suggest to keep VOLTHA separated by deploying them in different namespaces. For example to install a second VOLTHA stack do:
helm upgrade --install --create-namespace \
 -n voltha2 voltha2 onf/voltha-stack \
 --set global.stack_name=voltha2 \
 --set global.voltha_infra_name=voltha-infra \
 --set global.voltha_infra_namespace=infra

If you add a different number of ONOS you need also to tell the ofagent to connect to all of them by adding it to the voltha-stack command. The following is an example for 3 ONOS instances.

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra \
  --set voltha.onos_classic.replicas=3

To deploy a released version of the voltha-stack chart you can specify the --version in the command. As an example to deploy VOLTHA 2.8 LTS release you can use --version 2.8.0 like:

helm upgrade --install --create-namespace \
  -n voltha --version 2.8.0 voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra

Enable tracing in VOLTHA (optional)

To enable tracing across the VOLTHA components add --set global.tracing.enabled=true to the install command, for example:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra \
  --set global.tracing.enabled=true

NOTE that the Jaeger pod must be up and running before the VOLTHA stack starts in order for the components to register.

Enable log correlation in VOLTHA (optional)

To enable log correlation across the VOLTHA components add --set global.log_correlation.enabled=true to the install command, for example:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra \
  --set global.log_correlation.enabled=true

Use the OpenONU python adapter

Up to release 0.10.0 of the voltha-stack chart you can still use the openonu-py version of the adapter.

NOTE that this adapter is now unsupported, so you're in uncharted territory from now on.

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set global.voltha_infra_name=voltha-infra \
  --set global.voltha_infra_namespace=infra \
  --set voltha-adapter-openonu.use_openonu_adapter_go=false

Deploy BBSim

BBSim is a broadband simulator tool that is used as an OpenOLT compatible device in emulated environments.

In order to install a single BBSim instance to test VOLTHA, you can use the BBSim helm chart:

helm upgrade --install -n voltha bbsim0 onf/bbsim --set olt_id=10

While it's not mandatory to install BBSim in the same namespace as the VOLTHA stack it's advised to do so to make explicit which stack is controlling it.

Installing and Configuring `voltctl`

voltctl is a replacement for the voltha-cli container in VOLTHA that provides access to the VOLTHA CLI when a user connects to the container via SSH. voltctl provides a use model similar to docker, etcdctl, or kubectl for VOLTHA.

As voltctl is a binary executable as opposed to a Docker container it must be installed separately onto the machine(s) on which it is to be run. The Release Page for voltctl maintains of pre-built binaries that can be installed. The following is an example of how, in the example environment, voltctl can be installed with bash completion:

HOSTOS="$(uname -s | tr "[:upper:]" "[:lower:"])"
HOSTARCH="$(uname -m | tr "[:upper:]" "[:lower:"])"
if [ "$HOSTARCH" == "x86_64" ]; then
    HOSTARCH="amd64"
fi
sudo wget https://github.com/opencord/voltctl/releases/download/v1.7.4/voltctl-1.7.4-$HOSTOS-$HOSTARCH -O /usr/local/bin/voltctl
sudo chmod +x /usr/local/bin/voltctl
source <(voltctl completion bash)

If you are exposing the voltha-api service on 127.0.0.1:55555 as per the examples in this guide there is no need to configure voltctl, if you are exposing the service on a different port/IP you configure voltctl with:

voltctl -s <voltha-api-ip>:<voltha-api-port> config > $HOME/.volt/config

Deploying a different workflow

If you want to deploy VOLTHA with the appropriate configuration for the dt or tt worflow two example files are provided in the ./examples folder.

For you convenience here are the commands to deploy those workflows:

helm upgrade --install -n infra voltha-infra onf/voltha-infra -f examples/dt-values.yaml
helm upgrade --install -n voltha bbsim0 onf/bbsim --set olt_id=10 -f examples/dt-values.yaml
helm upgrade --install --create-namespace   -n voltha voltha onf/voltha-stack   --set global.stack_name=voltha   --set voltha_infra_name=voltha-infra   --set voltha_infra_namespace=infra

DT-FTTB

helm upgrade --install -n infra voltha-infra onf/voltha-infra -f examples/dt-fttb-values.yaml
helm upgrade --install -n voltha bbsim0 onf/bbsim --set olt_id=10 -f examples/dt-fttb-values.yaml
helm upgrade --install --create-namespace   -n voltha voltha onf/voltha-stack   --set global.stack_name=voltha   --set voltha_infra_name=voltha-infra   --set voltha_infra_namespace=infra

helm upgrade --install -n infra voltha-infra onf/voltha-infra -f examples/tt-values.yaml
helm upgrade --install -n voltha bbsim0 onf/bbsim --set olt_id=10 -f examples/tt-values.yaml
helm upgrade --install --create-namespace   -n voltha voltha onf/voltha-stack   --set global.stack_name=voltha   --set voltha_infra_name=voltha-infra   --set voltha_infra_namespace=infra

TT-MacLearner

helm upgrade --install -n infra voltha-infra onf/voltha-infra -f examples/tt-maclearner-values.yaml
helm upgrade --install -n voltha bbsim0 onf/bbsim --set olt_id=10 -f examples/tt-maclearner-values.yaml
helm upgrade --install --create-namespace   -n voltha voltha onf/voltha-stack   --set global.stack_name=voltha   --set voltha_infra_name=voltha-infra   --set voltha_infra_namespace=infra

Using an ingress controller

A process to expose the VOLTHA API external to the Kubernetes cluster was described using the port-forward option from the kubectl command line tool. This mechanism, while convenient, is not recommended for a production deployment. For a production deployment a Kubernetes Ingress Controller is recommended.

There are many choices when deploying an ingress controller including open source and commercial options. This document does not recommend or require a specific ingress controller, but it is important to note that VOLTHA has only been tested using the NGINX Ingress Controller and thus examples for Kubernetes manifests are offered in that context.

To deploy the NGINX ingress controller please see the instructions on the NGINX site. For instructions on deploying an ingress controller when using a kind cluster, as is common for VOLTHA development, please see the instructions on the kind site.

VOLTHA Helm Charts

The VOLTHA Helm charts were updated to deploy Ingress resources for the etcd and VOLTHA API services. By default these Ingress resources are disabled and not deployed.

By default the Ingress resources are defined without a host option. This is convenient when running a single stack but is not sufficient when running multiple stacks and enabling external access to both stacks through the ingress controller.

When using multiple stacks and an Ingress controller virtual hosts will be required to differentiate the stacks and direct API requests to the correct service. If --set voltha.ingress.enableVirtualHosts=true is used when installing the voltha stack then a virtual host will be defined using <stack-name>.voltha.local. In order to use this virtual host this virtual host name will need to resolve to the IP address of your Ingress controller.

To include the Ingress resources when deploying VOLTHA the following values need to be set:

When Installing Infra

--set etcd.ingress.enabled=true

When Installing the VOLTHA Stack

--set voltha.ingress.enabled=true

Usage with `voltctl`

Once the Ingress resources are defined voltctl should be able to access the VOLTHA deployment without having to run kubectl port-forward commands for each specific service. The configuration to voltctl to utilize the ingress controller can be specified either via the command line options or via the voltctl configuration file (default ~/.volt/config).

The important settings are

NOTE: Note hostname and port will vary depending on how you deploy as well as how the Ingress controller is accessed from outside the cluster. If you are deploying VOLTHA locally edit your /etc/hosts file adding at the end the same hostname you specified in the voltctl config, as an example 127.0.0.1 voltha.voltha.com

--server/server - the value of the external IP address of the ingress controller and the port on which it is listening (ex: localhost:443 or voltha.voltha.com:30474).
--kvstore/kvstore - the value of the external IP address of the ingress controller and the point on which it is listening (ex: localhost:443)
--tls/tls.useTls - indicates that TLS should be used when communicating through the ingress controller, which is required for the NGINX ingress controller (ex: true)

CLI example:

voltctl --server=voltha.voltha.com:443 --kvstore=localhost:443 --tls version

voltctl configuration file example:

server: voltha.example.com:443
kafka: localhost:443
kvstore: localhost:443
tls:
  useTls: true
  caCert: ""
  cert: ""
  key: ""
  verify: false
grpc:
  timeout: 5m0s
  maxCallRecvMsgSize: 4M
kvstoreconfig:
  timeout: 5s

Ingress Controller References

Guide to setting up ingress on a kind cluster.

Post installation

Ok, now I have VOLTHA installed and everything is running. What can I do with it?

Sanity Checks

As first make sure that all components are running correctly:

$ kubectl get pods --all-namespaces
NAMESPACE     NAME                                                  READY   STATUS      RESTARTS   AGE
infra         bbsim-sadis-server-6fcbdf9bd8-s7srz                   1/1     Running     0          14m
infra         elasticsearch-master-0                                1/1     Running     0          14m
infra         voltha-infra-etcd-0                                   1/1     Running     0          14m
infra         voltha-infra-fluentd-elasticsearch-9df8b              1/1     Running     0          14m
infra         voltha-infra-fluentd-elasticsearch-rgbvb              1/1     Running     0          14m
infra         voltha-infra-fluentd-elasticsearch-vfrcg              1/1     Running     0          14m
infra         voltha-infra-freeradius-7cbcdc66f-fhlt6               1/1     Running     0          14m
infra         voltha-infra-kafka-0                                  1/1     Running     0          14m
infra         voltha-infra-kibana-6cc8b8f779-7tlvp                  1/1     Running     0          14m
infra         voltha-infra-onos-classic-0                           1/1     Running     0          14m
infra         voltha-infra-voltha-infra-onos-config-loader-whdtz    0/1     Completed   3          14m
infra         voltha-infra-voltha-tracing-jaeger-7fffb6cdf6-l5r8s   1/1     Running     0          14m
infra         voltha-infra-zookeeper-0                              1/1     Running     0          14m
voltha       bbsim0-6f9584b4dd-txtj4                              1/1     Running     0          66s
voltha       voltha-voltha-adapter-openolt-5b5844b5b6-htlvp       1/1     Running     0          91s
voltha       voltha-voltha-adapter-openonu-85749df5fc-n5kdd       1/1     Running     0          91s
voltha       voltha-voltha-ofagent-5b5dc9b7b5-htxt6               1/1     Running     0          91s
voltha       voltha-voltha-rw-core-7d69cb4567-9cn2n               1/1     Running     0          91s

Note that is completely fine if the onos-config-loader pod restarts a few times, it is a job that loads configuration into ONOS and will fail until ONOS is ready to accept the configuration.

Once all the kubernetes pods are in Ready and Running state make sure the adapter registered with the core.

In order to use voltctl you need to expose the voltha-api service:

kubectl -n voltha port-forward --address 0.0.0.0 svc/voltha-voltha-api 55555

If you have deployed multiple stacks you need to change the port-forward command to connect to the stack you want to operate, eg:
kubectl -n voltha2 port-forward --address 0.0.0.0 svc/voltha2-voltha-api 55555

Once that is done you can query rw-core for a list of adapters:

$ voltctl adapter list
ID                     VENDOR              TYPE                 ENDPOINT                     VERSION            CURRENTREPLICA    TOTALREPLICAS    LASTCOMMUNICATION
brcm_openomci_onu_1    VOLTHA OpenONUGo    brcm_openomci_onu    voltha_brcm_openomci_onu    unknown-version    1                 1
openolt_1              VOLTHA OpenOLT      openolt              voltha_openolt              3.0.2              1                 1

Provisioning an OLT

Once you completed the Sanity Checks you can provision an OLT. We suggest to start with BBSim (see above for installation instructions).

To create and enable the OLT device in VOLTHA you can use these voltctl commands:

voltctl device create -t openolt -H bbsim0.voltha.svc:50060
voltctl device list --filter Type~openolt -q | xargs voltctl device enable

Once the OLT device is enabled you will see that an emulated ONU is reported to VOLTHA:

$ voltctl device list
ID                                      TYPE                 ROOT     PARENTID                                SERIALNUMBER    ADMINSTATE    OPERSTATUS    CONNECTSTATUS    REASON
42b04dfc-a253-46a0-8b96-da0551648fd5    brcm_openomci_onu    false    92a67593-06fa-4fad-87cb-b8befab90a56    BBSM00000001    ENABLED       ACTIVE        REACHABLE        initial-mib-downloaded
92a67593-06fa-4fad-87cb-b8befab90a56    openolt              true     b367cec6-a771-417a-94a7-8b8922fac587    BBSIM_OLT_0     ENABLED       ACTIVE        REACHABLE

Running the sanity test (optional)

If you want to run the sanity-test you can:

git clone https://github.com/opencord/voltha-system-tests.git && cd voltha-system-tests

export KUBECONFIG="$(k3d kubeconfig write voltha-dev)" # or you KUBECONFIG file
mkdir -p ~/.volt
voltctl config > ~/.volt/config
export VOLTCONFIG="~/.volt/config"
make sanity-kind

This assumes that both the onos-ssh, onos-rest and voltha-api ports are forwarded on the host and bbsim was installed with helm install -n voltha bbsim0 onf/bbsim --set olt_id=10.

Remove VOLTHA from your cluster

NOTE that this is not required as part of your development loop. In that case you should be able to simply upgrade the component you are working on (see next section).

If you need to completely uninstall everything that you installed following this guide, you can simply remove the installed helm charts:

helm del -n voltha voltha bbsim0
helm del -n infra voltha-infra

If you are using the bbsim-sadis-server component as well then remember to remove the kube-config configmap as well:

kubectl delete cm -n infra kube-config

Develop and test for VOLTHA

NOTE that this section is intended for development purposes, it's not a guide for in service software upgrade.

Develop with latest (master) code

The voltha-infra and voltha-stack charts pull the referenced version of every component, if your charts are up to date that will be the latest released one. If for your development and testing you'd like to have all the master of each component, independently if that has been officially tagged and release there is a provided dev-values.yaml. You can use it for the a voltha-stack like so:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set voltha_infra_name=voltha-infra \
  --set voltha_infra_namespace=infra \
  -f examples/dev-values.yaml

After the feature you are working on is released by modifying the VERSION file and removing -dev you can remove the dev-values.yaml file from your helm command.

Build a custom container

When you have made a change to a component in VOLTHA and you need to test it the first step is to build a custom docker image. Every VOLTHA component has a make target called docker-build that builds the docker image with the code in the local folder, e.g.:

DOCKER_REGISTRY="voltha-test/" DOCKER_TAG="custom-feature" make docker-build

this generates the docker image for that repo, e.g. voltha-test/voltha-openonu-adapter-go:custom-feature

(Up)load the custom container

Now that you have the custom image with your changes you have to make the image available to your k8s cluster.

If you are developing locally please load the image into your cluster of choice via the specific command (e.g. kind load docker-image).

If your cluster is remote, you want to share your image with others or test on Jenkins every VOLTHA component has a make target called docker-push that pushes the docker image to the registry you specify, e.g.

DOCKER_REGISTRY="voltha-test/" DOCKER_TAG="custom-feature" make docker-push

Upgrade a component or use a custom image

If you want to upgrade a component within a VOLTHA stack or use the custom built image you just created, you can use the same helm upgrade command as per the installation guide while providing a new image for one of the component, via custom helm flags. The following is an example for the openonu adapter go with the image we built during previous steps:

--set voltha-adapter-openonu.images.adapter_open_onu_go.repository=voltha-test/voltha-openonu-adapter-go \
--set voltha-adapter-openonu.images.adapter_open_onu_go.tag=custom-feature

A full command looks like the following:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set voltha_infra_name=voltha-infra \
  --set voltha_infra_namespace=infra \
  --set voltha-adapter-openonu.images.adapter_open_onu_go.repository=voltha-test/voltha-openonu-adapter-go \
  --set voltha-adapter-openonu.images.adapter_open_onu_go.tag=custom-feature

If as part of your development process you have published a new image with the same tag you can force its download simply by restarting the pod, for example:

kubectl delete pod -n voltha $(kubectl get pods -n voltha | grep openonu | awk '{print $1}')

In order for this to work the imagePullPolicy has to be set to Always.

Test changes to a chart

If you are working on an helm chart you can install a stack omitting that component, and then use the local copy of your chart to install it, eg:

helm upgrade --install --create-namespace \
  -n voltha voltha onf/voltha-stack \
  --set global.stack_name=voltha \
  --set voltha_infra_name=voltha-infra \
  --set voltha_infra_namespace=infra \
  --set voltha-adapter-openonu.enabled=false

helm upgrade --install --create-namespace \
  -n voltha opeonu-adapter voltha-adapter-openonu \
  --set global.stack_name=voltha \
  --set adapter_open_onu.kv_store_data_prefix=service/voltha/voltha_voltha \
  --set adapter_open_onu.topics.core_topic=voltha_voltha_rwcore \
  --set adapter_open_onu.topics.adapter_open_onu_topic=voltha_voltha_brcm_openomci_onu \
  --set services.kafka.adapter.service=voltha-infra-kafka.infra.svc \
  --set services.kafka.cluster.service=voltha-infra-kafka.infra.svc \
  --set services.etcd.service=voltha-infra-etcd.infra.svc

Known issues

Known VOLTHA issues are tracked in JIRA. Issues that may specifically be observed, or at the very least were discovered, in this environment can be found in JIRA via a JIRA Issue Search.

Pre-patchset submission Checks

On patchset submission, jobs are run in Jenkins that validate the correctness of the helm charts.

The code for these jobs can be found in helm-repo-tools

The two scripts that should be run to test are:

helmlint.sh
chart_version_check.sh