Airflow Kubernetes executor , Tasks in queued state and pod says invalid image and uses local executor - python-3.x

Question

strong textI have installed airflow in docker and using kubernetes executor unable to trigger dags. Dag runs are in queued state. KubernetesExecutor is creating pod but it says invalid image. If i describe the pod it uses local executor instead of kubernetes executor. Please help
Attaching log file for reference
**kubectl describe pod tablescreationschematablescreation-ecabd38a66664a33b6645a72ef056edc
Name: swedschematablescreationschematablescreation-ecabd38a66664a33b6645a72ef056edc
Namespace: default
Priority: 0
Node: 10.73.96.181
Start Time: Mon, 11 May 2020 18:22:15 +0530
Labels: airflow-worker=5888feda-6aee-49c8-a94b-39cbe5758062
airflow_version=1.10.10
dag_id=Swed-schema-tables-creation
execution_date=2020-05-11T12_52_09.829627_plus_00_00
kubernetes_executor=True
task_id=Schema_Tables_Creation
try_number=1
Annotations: <none>
Status: Pending
IP: 172.17.0.46
IPs:
IP: 172.17.0.46
Containers:
base:
Container ID:
Image: :
Image ID:
Port: <none>
Host Port: <none>
Command:
airflow
run
Swed-schema-tables-creation
Schema_Tables_Creation
2020-05-11T12:52:09.829627+00:00
--local
--pool
default_pool
-sd
/root/airflow/dags/User_Creation_dag.py
State: Waiting
Reason: InvalidImageName
Ready: False
Restart Count: 0
Environment:
AIRFLOW__CORE__EXECUTOR: LocalExecutor
AIRFLOW__CORE__SQL_ALCHEMY_CONN: postgresql://airflowkube:airflowkube#10.73.96.181:5434/airflowkube
Mounts:
/root/airflow/dags from airflow-dags (ro)
/root/airflow/logs from airflow-logs (rw)
/var/run/secrets/kubernetes.io/serviceaccount from default-token-64cxg (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
airflow-dags:
Type: HostPath (bare host directory volume)
Path: /data/Naveen/Airflow/dags
HostPathType:
airflow-logs:
Type: HostPath (bare host directory volume)
Path: /data/Naveen/Airflow/Logs
HostPathType:
default-token-64cxg:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-64cxg
Optional: false
QoS Class: BestEffort
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled <unknown> default-scheduler Successfully assigned default/swedschematablescreationschematablescreation-ecabd38a66664a33b6645a72ef056edc to evblfnclnullnull1538
Warning Failed 2m15s (x12 over 4m28s) kubelet, evblfnclnullnull1538 Error: InvalidImageName
Warning InspectFailed 2m (x13 over 4m28s) kubelet, evblfnclnullnull1538 Failed to apply default image tag ":": couldn't parse image reference ":": invalid reference format**

Answer 1

It looks like there is no image defined for the workers. You can set this in the airflow.cfg. Be sure to set these worker_container_repository and worker_container_tag. These are null by default which results in the behaviour you're experiencing.
A working example:
AIRFLOW__KUBERNETES__WORKER_CONTAINER_REPOSITORY: "apache/airflow"
AIRFLOW__KUBERNETES__WORKER_CONTAINER_TAG: "1.10.10-python3.6"
AIRFLOW__KUBERNETES__RUN_AS_USER: "50000"
The airflow docs have some more detailed info on the available properties: https://airflow.apache.org/docs/stable/configurations-ref.html#kubernetes