Skip to content

Latest commit

 

History

History
123 lines (89 loc) · 5.87 KB

README.md

File metadata and controls

123 lines (89 loc) · 5.87 KB
Raw

MicroFunctions Helm Chart

Tested with:

  • Kubernetes 1.18

  • Helm 3.5.2

  • Knative 0.21

Chart Details

This chart will do the following:

  • Deployment of an Nginx server that serves as storage frontend and hosts the web GUI

  • StatefulSet of 3 Riak nodes that serve as global data storage

  • Deployment of a standalone Elasticsearch server to collect tracing data

  • StatefulSet of caches to serve datalayer access locally

  • Start the Management MicroFunctions workflow as a Knative service (created through a job)

Installing the Chart

Our helm chart by default installs container images from a private registry. To build the components and push images to a private registry at localhost:5000 (use kube forwarder), use the following commands:

[user@vm knix]# make -C Sandbox push
[user@vm knix]# make -C ManagementService push
[user@vm knix]# make -C DataLayerService push
[user@vm knix]# make -C riak push
[user@vm knix]# make -C GUI push
[user@vm knix]# make -C TriggersFrontend push

The push target in deploy/helm/Makefile can be used to build and push all container images:

[user@vm knix]# make -C deploy/helm push

The REGISTRY environment variable can be used to have make tag and push container images to a custom registry, e.g. $ REGISTRY=my-registry.com make -C deploy/helm push or it can be modified in ./docker.mk.

To deploy the Helm chart with a release name mfn to the default namespace from the Helm chart sources, use:

[user@vm knix]# helm install --name mfn deploy/helm/microfunctions/

The deploy target in deploy/helm/Makefile can be used to build and push container images and deploy the Helm package in one step:

[user@vm knix]# cd deploy/helm
[user@vm knix]# make deploy

  • a namespace can be chosen with --namespace my-ns

  • the --debug option will provide detailed output of what is being deployed

  • The variable imageRepo can be set to pull containers from a custom registry, e.g. --set imageRepo=my-registry.com

  • The variables httpProxy and httpsProxy can be set to pass proxy environment variables to all workflow sandboxes run by the platform, e.g. --set httpProxy=http://my-proxy.corporate.com

  • Further variables concerning container resource requests and limits can be overwritten. Please see deploy/helm/microfunctions/values.yaml for more configuration options.

Connecting to the web GUI

Nginx is exposed through the Service nx- (e.g. nx-mfn) by default on a NodePort 32180 (optionally, nginx can be configured with an SSL port). There's no Ingress created by default.

Choosing a custom container repository

We're not hosting docker containers, so this Helm chart uses registry.kube-system.svc.cluster.local as the default container repository. Container images for each of the components are built and pushed individually and can be created globally from the source root:

[user@vm knix]# docker login my-registry.com
[user@vm knix]# REGISTRY=my-registry.com make push
[user@vm knix]# helm install --name mfn --set imageRepo=myregistry.com deploy/helm/MicroFunctions.tgz

Resource requirements

Component Pods Containers CPU
request / limit		 RAM
request / limit		 PersistentVolume
Deployment nx-mfn 1 nginx 0.1 / 0.5 1.0 / 1.0 0.0 / 0.0
Deployment tf-mfn 3 triggers-frontend 6.0 / 12.0 3.0 / 24.0 0.0 / 0.0
StatefulSet dl-mfn 3 datalayer 12.0 / 12.0 24.0 / 24.0 0.0 / 0.0
StatefulSet es-mfn 1 elasticsearch 2.0 / 4.0 2.0 / 4.0 0.0 / 0.0
StatefulSet rk-mfn 3 riak 12.0 / 12.0 24.0 / 24.0 0.0 / 0.0
Deployment wf-management 1 sandbox 2 / 4 2.0 / 4.0 0.0 / 0.0
Workflows 1 sandbox 1 / 1 1.0 / 2.0 0.0 / 0.0
disk

Resource requirements can be edited modifying the respective resources requests and limits in the Helm chart values:

  • manager.setup.resources controls the job that runs only once to start our management service as a KNIX workflow
  • manager.managementSandbox.resources configures the size of the management service's sandboxes
  • manager.sandbox.resources sets the default resources used with every workflow sandbox that is being deployed
  • datalayer.resources, riak.resources, nginx.resources and triggersFrontend.resources sets the resource request limits of those components, respectively

To resize the riak deployment, it's also crucial to adapt the Riak JVM and cache memory settings. Pay attention to these settings when resizing Riak pods:

riak:
  # Has to fit in the memory configuration alongside Riak (default setting claims half the pod memory)
  LevelDbMaximumMemory: 4294967296
  AntiEntropy: "passive"
  # Should be <= resources.limits.cpu
  ErlangSchedulersTotal: 4
  ErlangSchedulersOnline: 4