openshift-scale / images Goto Github PK

The current setup triggers an image build in quay if a new commit gets merged into the repo, we need a way to trigger it on a nightly basis to get the latest content including clusterloader binary and snafu into the image.

Noticing different outputs from oc client running inside the scale-ci-workload pod, vs. the one that was downloaded from the https://openshift-release-artifacts.svc.ci.openshift.org/4.2.0-rc.5/ , which is what my oc server version.

Output differences (also see below):
No resources found in cakephp-mysql0 namespace.
vs
No resources found.

Image ID: quay.io/openshift-scale/scale-ci-workload@sha256:bc2f318694b8006ae6656b6e041608579c00ed37d89623264e3a54fe169580c2

Before updating the oc client manually inside the workloads pod:

# oc version
Client Version: v4.2.0-alpha.0-198-g3780d2d
Server Version: 4.2.0-rc.5
Kubernetes Version: v1.14.6+2e5ed54
# oc  get pods --no-headers --show-labels -n cakephp-mysql0 -l deploymentconfig=cakephp-mysql-example
No resources found in cakephp-mysql0 namespace.

After updating the oc client manually inside the workloads pod:

# oc version
Client Version: openshift-clients-4.2.0-201910041700
Server Version: 4.2.0-rc.5
Kubernetes Version: v1.14.6+2e5ed54

# oc get pods --no-headers --show-labels -n cakephp-mysql0 -l deploymentconfig=cakephp-mysql-example
No resources found.

@sjug FYI

We should use the UBI base image for our container images whenever possible.

The golang version used to build the scraper binary as part of the scale-ci-workload image is hard coded, we need to modify it to use the latest version of golang available.

It is unclear if all packages in the scale-ci-workload container are required.

Double check the list here - https://github.com/openshift-scale/images/blob/master/scale-ci-workload/Dockerfile#L24

Anything that is not directly used in the workloads should be moved into a new container that we can add all "extra" tools to for debugging. This can then become the scale-ci-workload-debug container.

The network workload run fails. The pod log show the following:

pbench-uperf: python3 required, either directly or through SCL
[error][2019-11-22T12:56:56.608248986] Unable to verify connectivity and uperf installation on the following clients and servers: