More diagnostic information for Kubernetes deployments
long-term
A
Agreed Mockingbird
When a deployment is run and there is some issue with the containers coming up there is not much information in the pipeline execution to debug what happened.
It will be really helpful if we can include additional diagnostic information in the pipeline execution, for example the output of the following:
kubectl get pod <podName> -o=jsonpath='{.status.conditions}' | jq .
kubectl get pod <podName> -o=jsonpath='{.status.containerStatuses}' | jq .
Another example for helpful information is someoutput information like below:
kubectl get pod [pod]-o jsonpath='{range .status.conditions[*]}{.type}={.status}{"\t"}{.lastTransitionTime}{"\t"}{.reason}{"\t"}{.message}{"\n"}{end}'
once the timeout for k8s step hits, or maybe every 5 minutes to show the current statuses if it is taking too long.
Log In
Canny AI
Merged in a post:
Display error from Pod if Kubernetes deploy or migration fails
T
Theoretical Vole
If a Kubernetes deployment or migrations fails. Harness should display the log messages from the failing pod(s). This would accelerate the debugging of issues and would also stop people automatically assuming any errors are a Harness problem.
Canny AI
Merged in a post:
Provide additional information when the K8s error is "<service> exceeded its progress deadline"
X
Xylophone gold Caribou
When a deployment responds with this status it fails immediately. This error does not provide actionable data about what the actual issue is. If I described one of the pods or checked kubectl events, I'd see that another actual error caused the failure. Like the pod couldn't mount a volume or something else...
The users looking in Harness don't always have access to run the kubectl commands to get this data, so not having it available in the Harness console causes deploy teams to escalate to operations.
A
Aquamarine Minnow
Checking readiness gates is also useful
Rohan Gupta
Merged in a post:
Service and Error Logs from K8s Deployment available in Harness UI
C
Cheerful Whitefish
Greetings! Is there a way to see error logs of a failed Kubernetes deployment in the Harness UI? For example, if a deployment failed because the service itself had a config issue, we'd like to see the stdout logs from the service in the UI.
Rohan Gupta
long-term
We can provide more commands to run diagnostics. We will add this to our long term backlog.