-
Notifications
You must be signed in to change notification settings - Fork 2.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Proposal] Remove distros references from READMEs of individual components #30657
Comments
I think we started discussing this with #30164 I'll try and answer from my personal perspective, and I'm happy to talk about this at a SIG meeting.
Yes, from personal experience our users typically will look at contrib components and check if our distribution is listed.
I don't mind if we remove vendor distributions, that said the requirements for the distributions are not aligned with vendors. The only requirements are: the distribution must be open source and the distribution link point to a publicly accessible repository. That said, I'm not sure I understand this:
Do you mean here that the list miss some vendors? Or that vendors miss some of the components included in their distribution?
So vendors with more complete lists and better documentation will be elevated above others? Or maybe a vendor with a more comprehensive distribution will get more traction with end users? I sure hope that's not true! Our distribution has markedly less components than others, but we certainly try to make up for it with our easy to use installers and great documentation :) (sorry, free marketing doesn't just make itself)
We've certainly carried that burden and I've shouldered along the way, but it was a sweet journey.
Sounds like a nice enhancement idea, would you be open to filing an issue to add this feature? |
The original intent behind listing distros in the readme was to answer the questions of whether a component was in the core or the contrib distro. I agree that including vendor distributions has added unnecessary overhead and would be ok with keeping only the list of distros maintained by the project in the readme. That being said, maybe the easier thing is to generate a page on opentelemetry.io that lists the components of each distro we support. I don't have a strong preference here, but anecdotally, as a user, going to the component's readme tends to be my first choice for information, since it also includes the configuration I would need for that component. |
@codeboten And it's completely reasonable workflow (for config etc), but I have a hard time believing that you go there to decide which vendor distro to pick.
@atoulme This is a very different workflow. If your user already has your distro in mind, they can just as easily go to the distro docs and see which components are included.
I don't follow why this is even a question - are there planes to host multiple distros from contrib? Even if we did, the source of truth for what's included in the distro would be the distro's builder config, and if it's used to auto-gen the matrix in the readme I wouldn't have an issue with that. But the current mechanism, as I understand (automation or not) is the mapping must be maintained manually in each component, which makes no sense. It's like if Linux were tracking in its source code where it's being used. |
@yurishkuro right, this is more "i just found out about this component, and i want to use it in an otel distro, which one includes it" kinda question i want to answer |
I am sorry if I'm asking dumb questions. Thanks for bearing with me through it. |
I plan on following up on this today and open a PR to any other distribution that is not core and contrib. You raise additional concerns in this issue. I will follow up on them as well with additional issues. |
This issue has been inactive for 60 days. It will be closed in 60 days if there is no activity. To ping code owners by adding a component label, see Adding Labels via Comments, or if you are unsure of which component this issue relates to, please ping |
sorry, I didn't follow up on this. I'll add it to the SIG meeting after KubeCon EU to discuss. |
Personally, I like the idea of removing references to vendor distributions from the repo, and I'm the one who opened up the chore pull request linked in this issue. 👍🏼 Context: The reason I opened that pull request was because there was a Collector SIG a while back that ran distribution metrics and noticed our listed distro only had one component in it, which was unlikely (and incorrect). At the time, I was asked to make the updates listed in the pull request. We're adding in (and removing) components regularly in our distro. For example, we recently added some of the k8s receivers to our distro out of a need, and immediately removed the logging component as it became deprecated in favor of the debug component. Updating each of the components in contrib every time we do this certainly adds overhead for us, and for you. I'm a fan of removing this overhead. However, on the flip side, the specific component we have been contributing (which is still in development & not built into the contrib distro yet) made sense to have our distro listed on it because that's the only place it's currently built into and feature complete. It has taken time to contribute this component upstream because of the processes in place (not a knock on the processes, just a fact). Thus, if someone wants to use the component with all it's feature in a distro they can just deploy, it made sense (and I believe was the ask at the time) to have it listed in our distro through the metadata. That may be more of an edge case, but I do think it's an important example. In full transparency (as a distro maintainer) it absolutely is free marketing. But IMO, it's not without its rewards for both vendors and the project. It's been easy to search and filter who supports OTEL, and what they support. For example, I've looked at the components through the registry to make comparisons on ADOT vs OTEL Contrib distorts in the past, helping better inform antiquated enterprise architecture conversations. And I'm sure there's benefits (if albeit small and bespoke) from searchabililty on the internet and within git. I say that to say, there are some benefits to having it listed, and there was context why I made that original pull request, but I'm totally in favor of removing that overhead and finding a better, more minimal, solution. I also generally recommend most folks just build their own distro with the base set of components they need anyway as this follows the security principle of least privilege, using renovate to auto bump the latest from contrib. |
@adrielp I personally don't mind acknowledging the source of the contribution, e.g. "Developed by , included in the circa 2023", as a static notice that doesn't need to change. For up-to-date reference of distros that host the components I would invert the responsibilities and have the distro maintainers publish a standard file (possibly generated by ocb) and the OTEL repo would only store a static reference to those files. The OTEL website could periodically pull those files and rebuild a search index. |
@yurishkuro - I like & can support both of those solutions! |
Fixes #30657 This removes all distributions linked to this repository maintained outside of OpenTelemetry.
Related to open-telemetry/opentelemetry-collector-contrib#30657 This removes all distributions linked to this repository maintained outside of OpenTelemetry. Co-authored-by: Dmitrii Anoshin <anoshindx@gmail.com>
Component(s)
No response
Is your feature request related to a problem? Please describe.
As example, #29795 performs unnecessary maintenance work to list which components are included in a particular vendor distribution.
Describe the solution you'd like
Remove references to vendor distros from the repository.
I am ok with keeping one page that simply lists all distros that people bothered to add to that page. It will be low maintenance.
But the content of the distros should be documented in their respective websites (can be easily auto-generated if using
ocb
).Describe alternatives you've considered
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: