How to prevent Link component from calling SSR components with _rsc at the end of the URL?

[colonder]

Summary

I've been using Next's Link component all over my code, as per good practices and recommendations. I've also written all of page.tsx components to accommodate for SSR e.g. moved client-side components down the tree etc. to maximize performance. Unfortunately, as I noticed in production, and later read in the docs, the <Link/> component fetches a given page by adding _rsc=XXXXX at the end of the href, where XXX is some kind of id, regardless of whether it prefetches the page or not. This _rsc suffix tells Next.js to fetch a React Server Component optimized binary of the given page that was created during a production build. This behavior confuses Google Search Console greatly, as I have a lot of "duplicate page with the same canonical URL" entries like so:

https://www.domain.com/page1_rsc=XXXXX
https://www.domain.com/page2_rsc=XXXXX

where in reality their canonical URLs are just https://www.domain.com/page1 etc. This happens due to Google bot scanning pages, and following links like a normal browser would do. I couldn't find any clue on how to prevent this behavior, so I replaced all <Link/> components with a standard <a> tag. Is there any way to use <Link/> component, retain SSR, and get rid of this _rsc suffix? Or tell Google bot that this _rsc refers to the same component?

Additional information

No response

Example

No response

[Antonio112009]

I have the same issue. It is annoying that it appears in Google Search Console.
My Next.js version - 14.0.2-canary8

[tombennet]

But the requests for _rsc URLs while scrolling past link elements aren't really the issue here. Googlebot doesn't scroll, so these aren't a problem for SEO. The issue is that the initialCanonicalUrl prop in the JSON in the raw HTML source often contains an _rsc parameter. Once the page is rendered, the value of the canonical tag i correct, but Google is somehow discovering the _rsc versions regardless. We just need the _rsc to never be included in canonicals or href attributes of elements - that's enough for Google to not discover these variants.

[colonder]

You might be right, I'm not that proficient with Next.js. If you believe that's the core reason, maybe THAT could be opened as an issue?

[c0b41]

@tombennet do you have any example for Canonical Url problem, because i don't see anything like that?

[tombennet]

There's an issue related to initialCanonicalUrl here: #53274

It's not specifically related to the rsc param, but I suspect the underlying causes are related (i.e. non-canonical URLs in the initialCanonicalUrl prop). I'm still struggling to reliably reproduce the rsc appearance in initialCanonicalUrl - both Googlebot and my test crawls seems to encounter it only some of the time, which is driving me crazy.

[c0b41]

Well that issue more like search engine bot problem

[albing123]

Did you find any solution for this?
@colonder @Antonio112009

[colonder]

#58357 (reply in thread)

[david-uriostegui-a]

Did you find any solution for this? @colonder @Antonio112009

This should work
<Link prefetch={true}>

[david-uriostegui-a]

Or use this config

// next.config.js
module.exports = {
  experimental: {
    serverComponents: false, 
  },
};

[arfa123]

Or use this config

// next.config.js
module.exports = {
  experimental: {
    serverComponents: false, 
  },
};

There is no such option available in the Next.js configuration.

[kstroka]

Is there any updates on this issue? Did you solve that?

[icopp]

In short, you can't, because it's baked into the RSC behavior. There's pending work to refactor how that works from the React team to remove the ?rsc params/headers from how it works, but that hasn't been released yet.