litmuschaos / litmus-ansible Goto Github PK

• hosts: localhost
  become: no
  connection: local
  tasks:
  • name: "create temp file to upload S3"
    ansible.builtin.template:
    src: ./roles/k3s/node/templates/k3s.service.j2
    dest: ./roles/k3s/node/templates/k3s-node.service.j2
    mode: 0755

when run ansible-play book i got error

Unrecognized type <<class 'ansible.template.AnsibleUndefined'>> for ipwrap filter

What happened:
when we set LIB_IMAGE_PULL_POLICY policy to IfNotPresent, test case is always downloading the image.

What you expected to happen:
We want to skip the download part as image is available locally.

How to reproduce it (as minimally and precisely as possible):

1. Download all images locally.
2. Disable internet and then set LIB_IMAGE_PULL_POLICY to IfNotPresent
3. Run core DNS pod delete chaos test

Please parametrize path to tc-image.

The current implementation blocks normal use of network experiments on systems without internet access.
There is an ability to specify location of pumba image on a local repository but in the current state it will still try to download gaiadocker/iproute2 image from docker.io so that will cause an issue like following

kubectl logs pumba-netem-bmndgx -n test_ns
time="2020-07-20T11:11:42Z" level=error msg="failed to pull tc image" app=pumba error="Error response from daemon: Get https://registry-1.docker.io/v2/: read tcp x.x.x.x:x->x.x.x.x:x: read: connection reset by peer" function=github.com/sirupsen/logrus.WithError source="container/client.go:497"
time="2020-07-20T11:11:42Z" level=error msg="Error response from daemon: Get https://registry-1.docker.io/v2/: read tcp x.x.x.x:x->x.x.x.x:x: read: connection reset by peer" app=pumba function=github.com/alexei-led/pumba/pkg/container.dockerClient.NetemContainer source="container/client.go:240"
time="2020-07-20T11:11:42Z" level=error msg="failed to start netem for container" app=pumba error="Error response from daemon: Get https://registry-1.docker.io/v2/: read tcp x.x.x.x:x->x.x.x.x:x: read: connection reset by peer" function=github.com/sirupsen/logrus.WithError source="netem/netem.go:27"

Is this a BUG REPORT or FEATURE REQUEST?

FEATURE REQUEST

What happened:
Observed that current experiments are written with kubectl commands embedded in the shell commands. Recommend replacing it with Ansible modules for kubernetes. Eg: https://docs.ansible.com/ansible/latest/modules/k8s_module.html#synopsis
What you expected to happen:
If modules are embedded in the image, easier development process for new experiments as well as cleaner ansible code.
How to reproduce it (as minimally and precisely as possible):
Source Code analysis

Anything else we need to know?:
It is a recommendation. Would be happy to take up and convert the existing scripts to include modules and bake it into the image as well.

What happened:
CoreDNS pod delete experiment fails with litmus 1.13.8. I am using the litmuschaos/ansible-runner:1.13.8 image

What you expected to happen:
Experiment should run successfully.

How to reproduce it (as minimally and precisely as possible):
Use litmus 1.13.8 release and run CoreDNS pod delete experiment

Anything else we need to know?:
It fails because probeSuccessPercentage is not defined in the ansible_logic yaml file. Whereas, the generic pod delete experiment runs successfully as the variable is defined there.
Another thing is, community.kubernetes.k8 is depricated which is why it throws an error