# Calico Node Pods in CrashLoopBackOff: "Unable to get Typha service endpoints from Kubernetes"

## Problem

* The `calico-node` pod(s) are in `CrashLoopBackOff` state and/or constantly restarting.

{% tabs %}
{% tab title="Bash" %}

```bash
NAME                                              READY   STATUS             RESTARTS   AGE
calico-kube-controllers-f6b98779-qfmhh            1/1     Running            1          101d
calico-node-69nk4                                 0/1     CrashLoopBackOff   131        7h37m
calico-node-g7pz7                                 0/1     CrashLoopBackOff   131        7h37m
calico-node-mngnc                                 0/1     Running            130        7h37m
calico-node-p4ktd                                 0/1     CrashLoopBackOff   129        7h37m
calico-node-wlv9t                                 0/1     CrashLoopBackOff   130        7h37m
calico-node-zl5ng                                 0/1     CrashLoopBackOff   129        7h37m
calico-typha-95b7898f5-g559h                      1/1     Running            0          8h
calico-typha-95b7898f5-lmxbw                      1/1     Running            0          8h
calico-typha-95b7898f5-zmn2x                      1/1     Running            0          8h
```

{% endtab %}
{% endtabs %}

* The pod log (via `kubectl logs`) for the `calico-node` pod(s) shows the following error related to `calico-typha` .

{% tabs %}
{% tab title="Bash" %}

```bash
[ERROR][68] confd/discovery.go 126: Unable to get Typha service endpoints from Kubernetes. error=endpoints "calico-typha" not found
```

{% endtab %}
{% endtabs %}

* Similarly, the following error is recorded in the `calico-typha` pod(s).

{% tabs %}
{% tab title="Bash" %}

```bash
[ERROR][7] lookup.go 65: Failed to get Typha endpoint from Kubernetes error=endpoints "calico-typha" not found
[WARNING][7] rebalance.go 55: Failed to get number of Typhas error=<nil> numTyphas=0 thread="k8s-poll"
```

{% endtab %}
{% endtabs %}

* No `calico-typha` service (separate from the `DaemonSet` and `Deployment` objects) exists in the `kube-system` namespace.

## Environment

* Platform9 Managed Kubernetes – v5.5
* Calico

## Cause

The `calico-typha` service is missing from the `kube-system` namespace.

## Resolution

1. Create a file `calico-typha.yaml` with the following content.

{% tabs %}
{% tab title="Bash" %}

```bash
# Source: calico/templates/calico-typha.yaml
# This manifest creates a Service, which will be backed by Calico's Typha daemon.
# Typha sits in between Felix and the API server, reducing Calico's load on the API server.

apiVersion: v1
kind: Service
metadata:
  name: calico-typha
  namespace: kube-system
  labels:
    k8s-app: calico-typha
spec:
  ports:
    - port: 5473
      protocol: TCP
      targetPort: calico-typha
      name: calico-typha
  selector:
    k8s-app: calico-typha
```

{% endtab %}
{% endtabs %}

2. Apply the YAML to the affected cluster.

{% tabs %}
{% tab title="Bash" %}

```bash
kubectl apply -f calico-typha.yaml
```

{% endtab %}
{% endtabs %}

## Additional Information

* [Calico – Install Typha](https://projectcalico.docs.tigera.io/getting-started/kubernetes/hardway/install-typha)