Tracing tool to help troubleshooting #8361
Assignees
Labels
enhancement
This is the preferred way to describe new end-to-end features.
Comments
|
Thinking more on this, maybe think about a similar, or combined effort, to ensure other external_services are correctly configured. But also maybe that if connectivity is established, that can maybe somehow also ensure decent performance. Certainly just ensuring that users can quickly identify whether config is correct would be really valuable. I mention perf only because that's typically the next problem. They are connected but then experience timeouts etc. This all ties into stuff that @jmazzitelli and @nrfox are also doing around the improving how users can configure can configure and troubleshoot Kiali itself. |
Draft
9 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What do you want to improve?
The Tempo configuration introduced some complexity regarding the configuration of the tracing section in Kiali.
Trying to improve the situation with some documentation, this idea from @jshaughn about creating a tool that tries to autodiscover/validate the settings emerged.
What is the current behavior?
There is just a health check that it really doesn't help troubleshoot the problems.
What is the new behavior?
Create a tracing tool to correct the tracing settings for the Kiali configuration.
It can be an external tool like a hack script or, it can be integrated in Kiali.
The idea to be integrated in Kiali is because running as an external tool could introduce some additional issues related to the permissions, location, or the Kiali settings.
If the tool was an on-demand feature integrated in Kiali (For example, from the tracing node in the Mesh page) that based on the Kiali settings tries to perform some tests inside and outside the cluster (From the client side), and gave a report based on these tests.
For example:
the jaeger provider is configured but internal_url is pointing to the tempo api. Considerer change:
enternal_services.tracing.internal_url: "internal_tempo_url"
Or:
The external link is hidden because tempo is configured but Grafana is disabled. Consider change tempo_config.url_format to jaeger.
With some basic data, Kiali could try to connect using the http client or the grpc client, analyze the urls from the configuration, and try to infer the right urls based on some parameters, like http/grpc, the provider, with authentication, ...
The text was updated successfully, but these errors were encountered: