Question: What is Frisbee ?
Answer: Frisbee is a test-suite for Kubernetes.
Q: The service seems fine, but I get a Failed message.
A: The service run in a Pod that may host multiple containers. The application contrainer, the telemetry container, and so on. Given that, if the application seems fine, it is perhaps one of the sidecar containers that has failed.
Q: I changed some templates, but the changes does not seem to affect the Test Plan.
A: The changes are local and must be posted to the Kubernetes API. To update all templates within a chart use:
>> helm upgrade --install --wait my-example ./charts/example --debugQ: My experiment was running perfectly a few hours ago. Now, nothing works.
A: A possible explanation is that you do not specify the container version. If so, the latest version is retrieved for each run. And if there are incompatibilities between version, these incompatibilities will be reflected to your experiment.
Q: All I see in Grafana is a dot. There are no lines
A: This is likely to happen when the duration of the experiment is too short. In general, we use 1/4 resolution in order to make Grafana plots more readable. If you wish for a greater granularity, you can edit the chart and change resolution to 1/1.
Q: My plots in Grafana are not in line. The times are different
A: This is likely to happen if you have a change the resolution of one graph, without changing the other. Go and set the same resolution everywhere.
Q: I see dead plots ... on Grafana
A: Kubernetes v.1.22 drops support of cgroups v1 in favor of cgroup v2 API. If you run Kubernetes 1.22 make
sure that you have set systemd.unified_cgroup_hierarchy=1 in the grub configuration.
https://rootlesscontaine.rs/getting-started/common/cgroup2/
Q: When trying to install a Chart, I
get helm.go:81: [debug] found in Chart.yaml, but missing in charts/ directory:
A: This may happen if the dependencies of a chart (aka subcharts) are not installed. To fixed it, simply update the
chart. for the chart platform (aka Frisbee), do the following
>> helm dependency update charts/platform/**Q: Pods don't have internet access **
A: This may happen either because you have not enabled DNS on microk8s (microk8s enable dns) or because your
firewall is blocking DNS traffic (sudo ufw allow out to any port 53). If non of the above work, retry to reboot you
machine !
Disable your firewall
- DNS issues will manifest as a
[UFW BLOCK]...entries on thedmesg. - If you see that, disable your firewall via
sudo ufw disable
**Q: Ingress does not work **
A: Ingress issues will appear as 404 HTTP codes when trying to access the HTTP page.
If so, run kubectl describe ingress | grep "dashboard:" and access the HTTP page via ip:port scheme (e.g, 10.1.128.37:8443)
If the page is accessible then restart your machine and retry.