Experience post-migration from Svelte 3 to Svelte 5

[inspiredlabs]

WARNING 10GB PER-DAY DOWNLOAD USAGE

Obviously, I need to look into the underlying cause of this issue – but, it seems post-migration from Svelte 3 to Svelte 5 my website is consuming up 10GB per day in download usage.

This is very hard to fathom as the content of the site has hardly changed – although I can pin-point the spike after the day I use the automatic migration tooling.

Please let me know if this experience is something that others are aware of/not aware of.

Any thoughts or guidance welcome.

[inspiredlabs]

I have added a public repo here: https://github.com/inspiredlabs/sv-migrate-to-consume-bandwidth

[inspiredlabs]

Did you check all the $effect usage? In Vercel you can check your usage what routes etc (from: @~Scott, Sveltedon! WhatsApp group).

Yes. It doesn't make any sense to me though because it says the most demanding route is the WHO brochure PDF page (https://inspiredlabs.co.uk/cfe/cfe-brochure-2025.pdf), but it only accounts for a minuscule amount of bandwidth in comparison with the consumption rate.

[rmunn]

Is your site perhaps being scraped by a Web crawler or two? I can't imagine that the PDFs are very large, but there's also an .mp4 on the cfe page. How large is it? Also, since the cfe-brochure-2025.pdf is the first link on the page, I wonder if all downloads coming from that page are being misattributed to the first link by some bug in Vercel's analysis system. Not very likely, of course, but when you have a really weird result, you tend to examine every possiblity, even the unlikely ones.

At any rate, I don't see a robots.txt file in your source, so you should perhaps consider whether Web crawlers are consuming your bandwidth.

[rmunn]

The high bandwidth usage happening right after migrating would also make sense if it's Web crawlers, because any crawlers that indexed your site a long time ago probably noticed "Hey, the contents of the site have changed" (which they had, even if it's only the Javascript that was different — but the HTML was probably slightly different as well) and would start re-indexing your site. And in the absence of a robots.txt file, they would hit every URL, including any .mp4 files that are relatively large.

[inspiredlabs]

I wonder if all downloads coming from that page are being misattributed to the first link by some bug in Vercel's analysis system.

I don't have a Pro account so I can't Open in Observabilty, in light of this do you know what/where I can add my own crude inspector?

[inspiredlabs]

At any rate, I don't see a robots.txt

I didn't add one intentionally to https://github.com/inspiredlabs when I regressed back to Svelte 3, in order to monitor the /usage page on Vercel. It's dropped right back down to the normal rate again.

I'll add robots.txt to the Svelte 5 if I understand what/where I need to monitor in the absence of a Pro account.

[inspiredlabs]

@rmunn Thanks for getting back to me.