The new Beijing release scalability, resiliency, and manageablity are described here. These capabilities apply to the OOM/Kubernetes installation.
Installation
Follow the OOM installation instructions at
http://onap.readthedocs.io/en/latest/submodules/oom.git/docs/index.html
TLab specific Upgrade of Portal
Below are the notes specific for the TLab environment. If a change is made in OOM charts then the following needs to be completed to upgrade the running charts in
- Find Rancher IP in R_Control-Plane tenant - we use “onap_dev” key and “Ubuntu” user to SSH. For example: “ssh -i onap_dev ubuntu@192.168.xx.xxx”.
- Login as ubuntu, then run "sudo -i" to login as root. The “oom” git repo is in the rancher vm's root directory, under “/root/oom”.
- Edit portal files at /root/oom/kubernetes/portal....
When complete run the following from /root/oom/kubernetes dir
> make portal .... > make onap .... # This will do the upgrade: helm upgrade dev local/onap --namespace onap -f onap/resources/environments/disable-allcharts.yaml --set portal.enabled=true # When complete run the following to ensure its in the process of coming up or complete. kubectl get pods -n onap -o wide | grep portal
- Rancher gui is at 192.168.xx.xxx:8080
There is an interactive cli as well in the rancher gui where you can run kubectl commands. Below command will show the list of portal services.
kubectl get services --namespace=onap | grep portal
Accessing the ONAP Portal using OOM and a Kubernetes Cluster
If everything is successful, then to access Portal - http://onap.readthedocs.io/en/latest/submodules/oom.git/docs/oom_user_guide.html#accessing-the-onap-portal-using-oom-and-a-kubernetes-cluster
Overview of the running system
> kubectl get services --namespace=onap | grep portal portal-app LoadBalancer 10.43.141.57 10.0.0.8 8989:30215/TCP,8006:30213/TCP,8010:30214/TCP 5d portal-cassandra ClusterIP 10.43.158.145 <none> 9160/TCP,7000/TCP,7001/TCP,7199/TCP,9042/TCP 5d portal-db ClusterIP 10.43.192.65 <none> 3306/TCP 5d portal-sdk ClusterIP 10.43.24.82 <none> 8990/TCP 5d portal-widget ClusterIP 10.43.101.233 <none> 8082/TCP 5d portal-zookeeper ClusterIP 10.43.0.82 <none> 2181/TCP 5d
> kubectl get pods --all-namespaces | grep portal onap dev-portal-app-b8c6668d8-56bjb 2/2 Running 0 2m onap dev-portal-app-b8c6668d8-g6whb 2/2 Running 0 2m onap dev-portal-app-b8c6668d8-xshwg 2/2 Running 0 2m onap dev-portal-cassandra-5ddbc59ffd-qc6rp 1/1 Running 0 2m onap dev-portal-db-6d7fc58648-sp9sf 0/1 Running 0 2m onap dev-portal-sdk-868f696cd7-mnjxk 0/2 Init:0/1 0 2m onap dev-portal-widget-694c45b75f-nqdtt 0/1 Init:0/1 0 2m onap dev-portal-zookeeper-db466fc-kggsw 1/1 Running 0 2m
Healthchecks
Verify that the portal healtcheck passes by the robot framework:
portal-app Scaling
To scale a new portal-app, set the replica count appropriately.
In our tests below, we are going to work with the OOM portal component in isolation. In this exercise, we scale the portal-app with 2 new replicas.
The below code needs to be added to the integration-override.yaml file.
portal:
portal-app:
replicaCount: 3
Run below command from "/root” folder to do helm upgrade
helm upgrade -i dev local/onap -f integration-override.yaml
portal-app Resiliency
A portal-app container failure can be simulated by stopping the portal-app container. The kubernetes liveness operation will detect that the ports are down, inferring there's a problem with the service, and in turn, will restart the container.
Here is where one of the running instances was deleted. And another is starting.
Here the new instance has started and the old one is still terminating
After the deleted instance has terminated the 3 instances are all running normally.
During this time there was no issues with the Portal Website.
TODO: In the following releases, the portal-db, portal-widget needs to be tested with similar resiliency and scaling tests. The portal-cassandra and portal-zookeeper requires coordination with MUSIC team to test the resiliency as there are some plans to provide MUSIC as a service.
Sanity Tests
Check Portal UI and perform sanity tests.
- After 3 instances of Portal are up, edit IP in /etc/hosts file, and logon as demo user on http://portal.api.simpledemo.onap.org:30215/ONAPPORTAL/login.htm
- Then killed 1 instance, I am able to continue on Portal page seamlessly
- Another test on failover timing, when killed all 3 instances, the new Portal processes are coming up within 30 seconds
Troubleshoot
In case of failures, below commands may get handy for accessing the pods/logs and troubleshooting the issue.
To find all portal pods:
> kubectl get pods --all-namespaces | grep portal onap dev-portal-app-b8c6668d8-56bjb 2/2 Running 0 2m onap dev-portal-app-b8c6668d8-g6whb 2/2 Running 0 2m onap dev-portal-app-b8c6668d8-xshwg 2/2 Running 0 2m onap dev-portal-cassandra-5ddbc59ffd-qc6rp 1/1 Running 0 2m onap dev-portal-db-6d7fc58648-sp9sf 0/1 Running 0 2m onap dev-portal-sdk-868f696cd7-mnjxk 0/2 Init:0/1 0 2m onap dev-portal-widget-694c45b75f-nqdtt 0/1 Init:0/1 0 2m onap dev-portal-zookeeper-db466fc-kggsw 1/1 Running 0 2m
From above list, now to check DB logs:
> kubectl logs --namespace=onap dev-portal-db-6d7fc58648-sp9sf -c portal-db 2018-05-30 19:49:16 139875765802880 [Note] mysqld (mysqld 10.2.15-MariaDB-10.2.15+maria~jessie) starting as process 1 ... 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Mutexes and rw_locks use GCC atomic builtins 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Uses event mutexes 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Compressed tables use zlib 1.2.8 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Using Linux native AIO 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Number of pools: 1 2018-05-30 19:49:16 139875765802880 [Note] InnoDB: Using SSE2 crc32 instructions
To check portal-app logs:
> kubectl logs --namespace=onap dev-portal-app-b8c6668d8-56bjb -c portal-app /start-apache-tomcat.sh: option -i value is /start-apache-tomcat.sh: option -n value is /start-apache-tomcat.sh: values for IP (-i) and/or name (-n) are empty or short /start-apache-tomcat.sh: Starting server from /opt/apache-tomcat-8.0.37 30-May-2018 19:48:29.480 INFO [main] org.apache.catalina.startup.VersionLoggerListener.log Server version: Apache Tomcat/8.0.37 30-May-2018 19:48:29.482 INFO [main] org.apache.catalina.startup.VersionLoggerListener.log Server built: Sep 1 2016 10:01:52 UTC 30-May-2018 19:48:29.482 INFO [main] org.apache.catalina.startup.VersionLoggerListener.log Server number: 8.0.37.0
To get inside the portal-app docker and access the application logs:
> kubectl exec -it dev-portal-app-b8c6668d8-56bjb -n onap /bin/sh Defaulting container name to portal-app. Use 'kubectl describe pod/dev-portal-app-b8c6668d8-56bjb' to see all of the containers in this pod. / # cd /opt/apache-tomcat-8.0.37/logs/onapportal /opt/apache-tomcat-8.0.37/logs/onapportal # ls application.log debug.log metrics.log audit.log error.log
Portal’s Rocket Chat channel - http://onap-integration.eastus.cloudapp.azure.com:3000/channel/portal