Segmentation fault on SSR Astro 5 site

Support
1

  • I have an astro site, where a few pages are rendered on demand (ssr).

  • After a successful build, on the home page(which is ssr) I get an error:

    image
    image1368×547 24.6 KB

On visiting the logs, here is what I found:

Feb 22, 04:57:05 PM: RequestId: 652c6b54-8d04-424c-ad7b-d66bebd47569 Error: Runtime exited with error: signal: segmentation fault
Runtime.ExitError

  • After some quick tests, I discovered that if I go directly to another page’s URL (csr/ssr), the error doesn’t happen again, even if I go back to the homepage, which really isn’t a solution but maybe might be helpful.

  • I found another forum post that is similar to mine but for nextjs so I cant replicate the fix: Segmentation fault on NextJS 14/13.5 website

2

Hi @arndom, thanks for writing in.

Kindly try the suggestions below to see if your issue works.

  • Delete the edge functions folder in .netlify folder
    OR
  • Downgrade @astrojs/netlify from 5.3.3 to 5.3.2.

Write back if you tried any of the above and the problem persists.
Thanks.

3
  • I deleted the edge functions folder but that had no effect - locally it’s an empty folder
  • I have this: "@astrojs/netlify": "^6.2.0", so I still downgrade?
5

Hi @arndom, since the suggestions did not work, usually this is a type of out-of-memory error for the lambda function.

We don’t have any special way to gain insight into what a deployed function is doing. There is no “internal-to-Netlify” profiler or other way to see the internals of a function invocation.

For this reason, kindly add console.log() and console.time() calls to the code responsible for this route to reveal more about what the code is doing when the segmentation fault occurs.
Doing so may be the best way to reveal the cause of the segmentation fault.

6

Hi @clarnx, where should I put the .log and .time calls in the code?

7

It’s a index.astro file that is server-side rendered. I can’t figure out where to put the logging calls that would show the error triggered by the lambda when attempting to render the page.

8

Hi @arndom, thanks for trying out the suggestions.
At this point, the best you can do is turn off server-side rendering if you are not using it to see if it will help resolve the issue. I also suggest you take the time to go through your build logs to see if there is any indication that might be causing the error.

To turn off server-side rendering in Astro, you need to change the output configuration in your astro.config.mjs file. Here’s how you can do it:

  1. Open your astro.config.mjs file.
  2. Look for the output property.
  3. Set the output value to "static" instead of "server" or "hybrid".

Here’s an example of how your configuration should look:

import { defineConfig } from "astro/config";

export default defineConfig({
  output: "static",
  // other configuration options...
});

By setting output to "static", you’re telling Astro to generate static HTML files at build time instead of rendering pages on the server.

It’s important to note that turning off server-side rendering will disable any server-side features you might be using, such as dynamic routes or server-side API endpoints. Make sure your site doesn’t rely on any server-side functionality before making this change.

For more information on Astro’s rendering modes, you can refer to the Astro documentation on Static Site Generation (SSG).

Hopefully the above helps give you some insights.
Thanks.

9

Thanks.

I resolved it by converting the dynamic component to a server island rather than opting for server-side rendering for the whole page. This handled the suspected out-of-memory lambda function.

1 Like