Frederic Branczyk
Frederic Branczyk
Agreed, definitely, both are important and useful!
We're using the default size so I think we don't need to make this configurable yet. I believe we can check the size of the proto message before and after...
Ultimately though, I think we should stream the data, but that we should only do once we truly implement our own protocol/format.
Parca Agent amd64 and arm64 of v0.9.1 have `.eh_frame` sections: https://gist.github.com/brancz/b163d154e9f8179051ef457f0f4a9983
It just dawned on me, this is one of those features that we wanted to enable using relabeling. This actually turns out to be a feature that we haven't implemented/finished...
Agreed, this needs to be possible.
There will most likely be a provisioning API for each tenant to configure rules. cc @metalmatze @squat for this we will definitely need a database so sooner or later we...
> Changing undefined fields should cause generation error. It should cause a validation error, which it does at CI time, but we should have validations as part of the tooling...
https://github.com/kubernetes/kubernetes/issues/64830 basically alludes to the point that this is not a language problem, but no matter what generates the manifest, we always have to dry run them against a cluster...
Not disagreeing in a general case, but if the build and validation happen within the same process, which we're perfectly capable of doing, then what's the difference? :) I'm not...