This message was deleted.

To run another container (not as an init container) in a Flyte pod, add it to the

containers

list in your

V1PodSpec

. All containers listed in

containers

will run in parallel within the pod, while those in

init_containers

run sequentially before the main containers start. For example, to add a sidecar or secondary container, simply append another

V1Container

to the

containers

list, not `init_containers`—see the Flyte pod template documentation and example with multiple containers. Here's a minimal example:

from kubernetes.client import V1Container, V1PodSpec

pod_spec = V1PodSpec(
    containers=[
        V1Container(name="primary", image="your-primary-image"),
        V1Container(name="sidecar", image="your-sidecar-image"),
    ],
    # ... other pod spec fields ...
)

Would you like a step-by-step breakdown or a full Flyte task example? Sources: - Flyte pod template documentation - Example with multiple containers 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.

update my code with another container and provide @ancient-wolf-19325

Add another

V1Container

to the

containers

list in your

dynamic_pod_spec_sv

function. For example, to add a sidecar container named `"sidecar"`:

from kubernetes.client import V1Container, V1PodSpec, V1ResourceRequirements, V1VolumeMount, V1LocalObjectReference, V1SecurityContext, V1Volume, V1PersistentVolumeClaimVolumeSource, V1SecretVolumeSource

def dynamic_pod_spec_sv(claim_name: str, guid: str, network_topology: str) -> V1PodSpec:
    pod_spec = V1PodSpec(
        image_pull_secrets=[V1LocalObjectReference(name="te13535-artifactory-cred")],
        node_selector=RecomputeConfig.NODE_SELECTOR,
        host_ipc=True,
        restart_policy="OnFailure",
        service_account=RecomputeConfig.SERVICE_ACCOUNT,
        service_account_name=RecomputeConfig.SERVICE_ACCOUNT,
        containers=[
            V1Container(
                name="sessionvalidator",
                image=RecomputeConfig.SESSION_VALIDATOR_IMAGE,
                image_pull_policy="Always",
                security_context=V1SecurityContext(privileged=True),
                resources=V1ResourceRequirements(requests={"cpu": "10", "memory": "10Gi"}),
                volume_mounts=[V1VolumeMount(name="gcs-volume", mount_path=RecomputeConfig.GCS_FOLDER)],
            ),
            V1Container(
                name="sidecar",
                image="alpine",
                command=["/bin/sh"],
                args=["-c", "echo sidecar running; sleep 1000"],
                resources=V1ResourceRequirements(requests={"cpu": "1", "memory": "100Mi"}),
                volume_mounts=[V1VolumeMount(name="gcs-volume", mount_path=RecomputeConfig.GCS_FOLDER)],
            ),
        ],
        init_containers=[
            V1Container(
                name="drivedata",
                image=RecomputeConfig.GCP_INIT_IMAGE,
                image_pull_policy="Always",
                command=["/bin/sh", "-c"],
                args=["sleep 100"],
                resources=V1ResourceRequirements(requests={"cpu": "10", "memory": "40Gi"}),
                volume_mounts=[V1VolumeMount(name="gcs-volume", mount_path=RecomputeConfig.GCS_FOLDER)],
            ),
        ],
        volumes=[
            V1Volume(
                name="gcs-volume",
                persistent_volume_claim=V1PersistentVolumeClaimVolumeSource(claim_name=claim_name),
            ),
            V1Volume(
                name="gitlab-secrets",
                secret=V1SecretVolumeSource(secret_name=RecomputeConfig.GITLAB_CERTIFICATES),
            ),
            V1Volume(
                name="gitlab-token-secret",
                secret=V1SecretVolumeSource(secret_name=RecomputeConfig.GITLAB_TOKEN),
            ),
        ],
    )
    return pod_spec

when i am keeping sidecar container it is went queued state @ancient-wolf-19325

If your Flyte task stays in the "queued" state after adding a sidecar container, it usually means the pod cannot be scheduled or started by Kubernetes. Common causes include resource requests exceeding available cluster resources, missing images, or misconfiguration in the pod spec. Check the pod's events and status using

kubectl describe pod <pod-name>

for specific errors, such as "Insufficient CPU/Memory" or "ImagePullBackOff" (Flyte pod template docs). Would you like step-by-step troubleshooting instructions for diagnosing the queued state? Sources: - Flyte pod template docs 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.

it is not due to resource issue, it is something elese in flyte ui it showing queued @ancient-wolf-19325

If the Flyte UI shows your task as "queued" and it's not a resource issue, it may be due to pod spec misconfiguration, such as missing or incorrect

primary_container_name

, image issues, or unsupported pod spec fields. Flyte requires the

primary_container_name

in the pod template to match a container in the

containers

list, and the primary container must be present and correctly configured (Flyte pod template docs). Check that: - The

primary_container_name

exactly matches one of the container names in

containers

. - All required fields (like images) are set and valid. - No unsupported or misspelled fields are present. Would you like a checklist or example for verifying your pod spec? Sources: - https://github.com/flyteorg/flyte/blob/master/flyteplugins/go/tasks/plugins/k8s/pod/plugin.go 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.

Hi @ancient-wolf-19325 can you erase all the messages

I'm here to help with questions about Flyte! I can't erase previous messages, but if you have a Flyte-related question or need clarification on something, please let me know how I can assist you. 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.