The "K8s Connector" allows you to programmatically access your Google Kubernetes Engine (GKE) cluster from your local environment. It leverages the google/cloud-sdk docker image and provides a container class with a range of functions that utilize kubectl commands.
Note
The library is mainly for accessing basic pods information, but not for performing any operations on resources to manage the cluster
- Collect pods status and logs upon receiving a 5xx response in automated API tests
- Check pods status as part of a test setup and skip the test if any pods are in
CrashLoopBackOffstatus - Check ongoing app deployment as a pre-hook for an API request and wait for it to complete before making the call
- Check pods status afrter a production release to ensure successful deployment
- Python 3.11+
- Docker and
google/cloud-sdkimage - A working Google Kubernetes Engine (GKE) cluster
- Service Account file with sufficient read permissions
- Create a new virtual env
- Run
pip install -e .
>>> from k8s_connector import K8sConnector
>>> k8s = K8sConnector(service_account_file_path="/path/key.json", cluster_name="my-cluster", zone="us-west1-a")
2024-01-01T00:00:00.000-0000 - Starting a container google/cloud-sdk:latest
2024-01-01T00:00:00.208-0000 - Started: c9b66e7152bbebe86445aaa9c6de4a2ee55b0ec8c48aa0b17a10d1ba8fcab7b6
2024-01-01T00:00:00.213-0000 - Activating Service Account credentials...
2024-01-01T00:00:00.213-0000 - Executing command: sh -c 'gcloud auth activate-service-account --key-file=/tmp/key.json --project myproject'
2024-01-01T00:00:01.133-0000 - output:
Activated service account credentials for: [[email protected]]
2024-01-01T00:00:01.158-0000 - Fetching credentials for the cluster...
2024-01-01T00:00:01.158-0000 - Executing command: sh -c 'gcloud container clusters get-credentials my-cluster --zone us-west1-a'
2024-01-01T00:00:02.660-0000 - output:
Fetching cluster endpoint and auth data.
kubeconfig entry generated for my-cluster.>>> exit_code, output = k8s.exec_run('kubectl version --client')
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl version --client'
2024-01-01T00:00:00.304-0000 - output:
Client Version: v1.28.12-dispatcher
Kustomize Version: v5.0.4-0.20230601165947-6ce0bf390ce3# Get pods
>>> pods = k8s.get_pods()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl get pods -A'
2024-01-01T00:00:00.879-0000 - output:
NAMESPACE NAME READY STATUS RESTARTS AGE
example app1-c44949f4ba-zsxpd 3/3 Running 0 30m
example app1-c44949f4ba-gjepp 3/3 Running 0 31m
example app2-685dbfc75d-fzcmr 3/3 Running 0 31m
example app2-685dbfc75d-pxhmj 3/3 Running 0 29m
...
# Get namespaces
>>> namespaces = k8s.get_namespaces()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl get namespaces'
2024-01-01T00:00:00.279-0000 - output:
NAME STATUS AGE
example Active 30d
...# Get pods
>>> pods = k8s.app1.get_pods()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl get pods -l app=app1 -n example'
2024-01-01T00:00:00.625-0000 - output:
NAME READY STATUS RESTARTS AGE
app1-c44949f4ba-zsxpd 3/3 Running 0 3d5h
app1-c44949f4ba-gjepp 3/3 Running 0 3d5h
...
# Get pods logs
>>> logs = k8s.app1.get_logs()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl logs -l app=app1 -n example --since=30s --tail=-1'
2024-01-01T00:00:01.342-0000 - output:
[2024-01-01T00:00:00.000000+00:00] [INFO]{fb0f6a7a-66e7-4411-802b-e9b501f196f7}[app1.api.foo] - GET /foo
[2024-01-01T00:00:00.000000+00:00] [INFO]{941e391a-3b0f-464b-a35b-979c1f084443}[app1.api.bar] - GET /bar
...
# Stream pods logs (with filtering)
>> k8s.app1.get_logs(follow=True, grep="ERROR|CRITICAL")
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl logs -l app=app1 -n example --since=30s --tail=-1 -f --max-log-requests=30 | GREP_COLOR="1;32" stdbuf -o0 grep -E -E "ERROR|CRITICAL" --color=always'
[2024-01-01T00:00:00.000000+00:00] [ERROR]{b9e75fd4-7f82-446e-92fd-48f23a81dbb4}[app1.api.foo] - foo
[2024-01-01T00:00:00.000000+00:00] [ERROR]{7e6f3618-2ec5-4fe5-ae93-4aaa387f3263}[app1.api.bar] - bar
...
# Get environment variables set as a configmap
>>> configmap_data = k8s.app1.get_configmap_data()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl get configmap -l app=app1 -n example -o json | jq '"'"'.items[] | select(.kind == "ConfigMap") | .data'"'"''
2024-01-01T00:00:00.609-0000 - output:
{
"APP1_ENV_VAR1": "true",
"APP1_ENV_VAR2": "false",
...
}
# Get metrics
>>> output = k8s.app1.top()
2024-01-01T00:00:00.000-0000 - Executing command: sh -c 'kubectl top pod --containers --sort-by=cpu -l app=app1 -n example'
2024-01-01T00:00:00.590-0000 - output:
POD NAME CPU(cores) MEMORY(bytes)
app1-c44949f4ba-zsxpd app1 617m 1253Mi
app1-c44949f4ba-gjepp app1 607m 1174Mi
...
# Wait for app deployment to complete after a successful build
>>> k8s.app1.wait_for_deployment()
2024-01-01T01:00:00.000-0000 - Waiting for app1 deployment to start...
2024-01-01T01:01:57.730-0000 - app1 deployment started
2024-01-01T01:01:57.730-0000 - Waiting for deployment of app1 to complete...
2024-01-01T01:01:57.730-0000 - Executing command: sh -c 'kubectl rollout status deployment/app1 -n example --timeout=360s'
2024-01-01T01:04:22.022-0000 - output:
Waiting for deployment "app1" rollout to finish: 2 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 2 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 2 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 2 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 3 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 3 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 3 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 3 out of 4 new replicas have been updated...
Waiting for deployment "app1" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "app1" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "app1" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "app1" rollout to finish: 3 of 4 updated replicas are available...
deployment "app1" successfully rolled out
2024-01-01T01:04:22.022-0000 - Executing command: sh -c 'kubectl get pods -l app=app1 -n example'
2024-01-01T01:09:22.657-0000 - output:
NAME READY STATUS RESTARTS AGE
app1-7d575bb744-7c2fg 3/3 Running 0 91s
app1-7d575bb744-9jxtc 3/3 Running 0 2m37s
app1-7d575bb744-fzcmr 3/3 Running 0 2m38s
app1-7d575bb744-pxhmj 3/3 Running 0 37s
2024-01-01T01:04:22.657-0000 - Deployment of app1 completed (took 144.926855802536 seconds)Tip
Most functions return the raw output by default, but some support parse parameter which will parse the output and return it as an object (eg. Returns the tabulated output as a list of dictionaries)