Skip to content

Latest commit

 

History

History
207 lines (138 loc) · 6.83 KB

gcp.md

File metadata and controls

207 lines (138 loc) · 6.83 KB
Raw

GCP - Google Cloud Platform

DevOps Bash tools for GCP, GKE, GCE etc

HariSekhon/DevOps-Bash-tools

Readme Card

GCE - Google Compute Engine

VM Instance Sizes

https://gcpinstances.doit.com/

Even more useful, you can click on a machine type and more details:

https://gcloud-compute.com/instances.html

VM Zones Gotcha

Pay special attention to the price per region table further down in a machine specifics page on gcloud-compute.com where you can see #Zones. If it says 2 zones, beware that your fancy Terraform 3 zone code, for example GKE clusters, is going to fail to create the VMs because Google haven't built that spec out in the 3rd zone yet, especially if you're deploying into one of the non-primary regions like europe-west-2 (London, UK). You will be forced to change the machine-type and redeploy.

Install GCloud SDK CLI

Follow the install doc or paste this to run an automated install script which auto-detects and handles Mac or Linux:

git clone https://github.com/HariSekhon/DevOps-Bash-tools
bash-tools/install/install_gcloud_sdk.sh

Initialize your config and authenticate, following the prompts:

gcloud init

Add SSH key to project

gcloud compute os-login ssh-keys add --key-file="$HOME/.ssh/id_rsa.pub"

If you're struggling to log in check your username eg. [email protected] instead of [email protected].

Add SSH key to specific VM

Since the metadata SSH needs to be in the format:

<username>:<ssh_key>
export VM=server1
export SSH_GCP_USERNAME=harisekhon
export SSH_KEY_PUB="$HOME/.ssh/id_rsa.pub"
gcloud compute instances add-metadata "$VM" --metadata-from-file ssh-keys=<(echo -n "$SSH_GCP_USERNAME:"; cat "$SSH_KEY_PUB")

You can iterate this using a script like gce_foreach_vm.sh in the DevOps-Bash-tools repo which has a regex filter for a subset of VMs if you only want to grant access to that subset.

Otherwise use the project wide SSH keys above.

Check you can see it under metadata ssh-keys

gcloud compute instances describe "$VM"

SSH Tunnelling Through a Bastion Host

See SSH Tunnelling

Set up access to GKE - Google Kubernetes Engine

First set up your GCloud SDK CLI as above.

Run the gke_kube_creds.sh script from the DevOps-Bash-tools repo's gcp/ directory.

This will find and configure all your kubernetes clusters in the current project.

gke_kube_creds.sh
kubectl config get-contexts

switch to the cluster you want:

kubectl config use-context <name>
kubectl get pods --all-namespaces

Then see Kubernetes for configs, scripts and .envrc.

Readme Card

See all the details you can query for a VM

See gcloud topic filters for the details on the --filter matching.

Prefer regex, it's the sharpest most accurate and flexible, but make sure it's anchored to not match other nodes eg. node1 should not match node10.

gcloud compute instances list --filter="name ~ ^${VM_NAME}$" --format=text

Find the field that contains the IP address:

gcloud compute instances list --filter="name ~ ^${VM_NAME}$" --format=text | grep -i ip

Get the IP Address of a specific VM

Use this if you are running a script like a Solr create collections against the IP address of a Solr node in the SolrCloud cluster.

gcloud compute instances list --filter="name ~ ^${VM_NAME}$" --format='get(networkInterfaces[0].networkIP)'

Get the names + IPs of all or a selection of VMs by regex name match

Clone DevOps-Bash-tools, then:

gcp/gce_host_ips.sh <optional_regex>

Get the IP of a random node in a cluster

Useful if you're running curl commands against an Elasticsearch or SolrCloud cluster.

gcloud compute instances list --filter="name ~ ^${VM_NAME_PATTERN}$" --format='get(networkInterfaces[0].networkIP)' | shuf | head -n1

Get the IP address of a Load Balancer

Useful to quickly get to an internal named load balancer by IP address to jump to the UI of an Elasticsearch or SolrCloud cluster.

gcloud compute forwarding-rules list --filter="name ~ ^${LOAD_BALANCER_NAME}$" --format='value(IPAddress)'

Get the IP of your Google FileStore NFS server

Quickly compare this to your config such as your Jenkins JCasC config per environment to ensure your config is pointing to the right IP

Notice the filestore name is in format projects/<PROJECT_ID>/locations/europe-west2-b/instances/<NAME> so we match the suffix /${NAME}

gcloud filestore instances list --filter="name ~ /${FILESTORE_NAME}$" --format='value(networks[0].ipAddresses[0])'

Enable APIs

gcloud services list --available | grep compute
gcloud services enable compute.googleapis.com

Partial port from private Knowledge Base page 2015+