Skip to content

Latest commit

 

History

History
88 lines (61 loc) · 2.92 KB

gke-hosted-kubernetes.md

File metadata and controls

88 lines (61 loc) · 2.92 KB
Raw

GKE (Google Kubernetes Engine)

Running on GKE inside GCP (Google Cloud Platform)

Docker Container

The GCP SDK is already installed inside the docker container docker run -ti quorumengineering/qubernetes

Prerequisites

  1. GCP account: https://cloud.google.com/free/
  2. GCP SDK: https://cloud.google.com/sdk/docs/quickstart-macos

Enable Kubernetes Engine

Creating a Kubernetes Cluster with GCP SDK

## select an existing project and zone.
$> gcloud init  

$> gcloud config list project

$> gcloud container clusters list

# all lowercase for name
$> gcloud container clusters create q1
...
NAME       LOCATION    MASTER_VERSION  MASTER_IP     MACHINE_TYPE   NODE_VERSION   NUM_NODES  STATUS
q1         us-east1-b  1.12.8-gke.10   34.73.114.36  n1-standard-1  1.12.8-gke.10  3          RUNNING

# or with a default region
$> gcloud container clusters create q1 --region=us-east1 

# or with default zone
$> gcloud container clusters create q1 --zone=us-east1-b 
NAME  LOCATION    MASTER_VERSION  MASTER_IP     MACHINE_TYPE   NODE_VERSION    NUM_NODES  STATUS
q1    us-east1-b  1.13.11-gke.14  35.231.41.38  n1-standard-1  1.13.11-gke.14  3          RUNNING

# delete cluster (include zone or region)
$> gcloud container clusters delete q1 --zone=us-east1-b

Setting up kubectl

> gcloud container clusters list
NAME  LOCATION    MASTER_VERSION  MASTER_IP     MACHINE_TYPE   NODE_VERSION    NUM_NODES  STATUS
q1    us-east1-b  1.13.11-gke.14  35.231.41.38  n1-standard-1  1.13.11-gke.14  3          RUNNING

# this will set the context and access enbabling the kubectl cmd to work with the GKE cluster
> gcloud container clusters get-credentials q1 --zone=us-east1-b

> kubectl get pods
No resources found.

# deploy a quorum network to the GKE cluster
> kubectl apply -f 7nodes/istanbul-7nodes-tessera/k8s-yaml-pvc

Troubleshooting

  1. pods are crashing right away Init:CrashLoopBackOff.
  • Are persistent volume mounts (PVM) being used? This issue might be due to hostPath as the storage backend, and redeploying the cluster. Try using PVM instead.

Revert to Previous gcloud Versions

$> gcloud components update --version 261.0.0

Status

$> gcloud container clusters list
NAME  LOCATION    MASTER_VERSION  MASTER_IP     MACHINE_TYPE   NODE_VERSION    NUM_NODES  STATUS
q1    us-east1-b  1.13.11-gke.14  35.231.41.38  n1-standard-1  1.13.11-gke.14  3          RUNNING

$> gcloud container clusters describe q1 --zone=us-east1-b

Delete hostpath

  • When using hostpath as the storage, delete the cluster and recreate it before deploying another cluster - this ensures that the underlying storage has been deleted as well.
  • gcloud container clusters delete $NAME