HTTP Collector Documentation Confusing

[kireledan]

After reviewing the HTTP Collector docs @ https://troubleshoot.sh/docs/collect/http/ , the page mentions that the collector can run inside a container.

"The http collector can be used to execute http requests from inside the cluster. "

However, after trying to test this on a personal cluster, I found that its unable to hit endpoints that a cluster can reach. It is able to hit local endpoints on my machine though.

Is there an undocumented flag for getting the HTTP request to run in the targeted cluster? Or is this just mis-worded / a planned feature.

[adamancini]

@kireledan I'll take a look, but can you share your test procedure showing how the HTTP collector failed in your case?

[adamancini]

taking a look at the code, the collector doesn't run inside a container, but executed by the support-bundle or preflight binaries, so I can see how it would fail to resolve outside the cluster - after talking with some of the engineers at Replicated this might not be the only collector that acts that way. I'll take a look for any more.

I think we should probably fix that so that the collectors default to run inside the cluster, and we should consider having a client-side run as an option so the collectors can cover both use cases.