Skip to content

Metadata API in destination service not configured for k8s Jobs #11531

New issue
New issue
Closed
#11543
#11541
Closed
Metadata API in destination service not configured for k8s Jobs#11531
#11543
#11541
Labels
bug
@siggy

Description

@siggy
siggy
opened on Oct 24, 2023

What is the issue?

I'm observing lots of failed to retrieve job from indexer warnings in the destination container logs, for pods belonging to Jobs:

linkerd2/controller/k8s/metadata_api.go

Lines 230 to 243 in 225ce93

case "Job":
parentObj, err = api.getByNamespace(Job, pod.Namespace, parent.Name)
if err != nil {
log.Warnf("failed to retrieve job from indexer %s/%s: %s", pod.Namespace, parent.Name, err)
if retry {
parentObj, err = api.client.
Resource(batchv1.SchemeGroupVersion.WithResource("jobs")).
Namespace(pod.Namespace).
Get(ctx, parent.Name, metav1.GetOptions{})
if err != nil {
log.Warnf("failed to retrieve job from direct API call %s/%s: %s", pod.Namespace, parent.Name, err)
}
}
}

It appears the metadata API client is only configured for Nodes and ReplicaSets:

linkerd2/controller/cmd/destination/main.go

Line 120 in 225ce93

metadataAPI, err := k8s.InitializeMetadataAPI(*kubeConfigPath, "local", k8s.Node, k8s.RS)

How can it be reproduced?

Start some k8s Jobs (I don't have a repro)?

Logs, error output, etc

seeing these warnings in the destination container:

2023-10-24T02:58:13.881735669Z time="2023-10-24T02:58:13Z" level=warning msg="failed to retrieve job from indexer [namespace]/[pod]: metadata informer (6) not configured"

output of linkerd check -o short

N/A

Environment

stable-2.14.1

Possible solution

Modify the call to InitializeMetadataAPI to include Job.

Additional context

No response

Would you like to work on fixing this bug?

maybe

Metadata

Metadata

Assignees

No one assigned

    Labels

    bug

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions