Can you share the deployment URL?

I had the same issue on my own app.

I was handing out invite links for people to claim a spot. Links weren’t one-time — they kept working after I thought I was done. Then I moved hosts and rotated the signing secret, and suddenly a bunch of old links died. I spent forever thinking the product broke.

What I finally got: expire and revoke aren’t the same thing. Expire is the clock on the link. Revoke is “kill this one now.” If the invite is just a signed URL with no row in a table, you can’t yank one link — you either wait for it to expire or rotate the secret and nuke every outstanding invite.

So for Railway you don’t need to recreate the project. That’s the nuclear option.

1. Project → Members (or workspace gear → People).
2. Remove the pending invite / that person from the list. That kills the old path.
3. Mint a new invite (email or Invite by link) and send that.
4. If they already joined on the old link, remove them as Admin, then re-invite if you still want them.

Fresh invite after remove. Don’t delete the project just to revoke a share link.

Sorry I can't share the url publicly. Can I share it directly with support team ?

good answer lol

You can open a private thread to talk directly with the team

Is this issue happening only to your users or also happened to you? If you're facing this issue yourself, You can try change your DNS to 1.1.1.1 (cloudflare) or 8.8.8.8 (Google)

If changing DNS works, then yes, the ISP may block Railway's network

One more thing on why that works —

Removing the pending invite is what actually revokes it. The old link can’t finish joining without that pending seat still sitting there. That’s the kill, not deleting the project.

They don't say anything about invite link..

My bad — I didn’t clarify myself.

I was thinking of something similar I hit on my own stuff: some users “couldn’t join / load” while the host still pinged. Turned out it wasn’t invites — it was DNS / wrong host / edge path. Marketing domain vs the real Railway app URL, or an ISP/CDN path dying while ping still looked fine. I jumped too fast and mixed that in without saying so.

For your case I’d still do what the mod said: private thread with the deploy URL, and have a broken user try 1.1.1.1 or 8.8.8.8. If that fixes it, it’s ISP DNS filtering Railway. If not, Railway can check the URL private. Sorry for the noise.

When I see someone going through something, even if it’s not the same code or stack, I map it to a problem I’ve already hit and figured out. I’m not trying to feed wrong info — I’m trying to offer how I got around something similar, so maybe they go “oh I didn’t think of that, maybe I could try it that way.”

I didn’t say that. Next time I’ll mark it as an example from my side when it’s not the exact same thing. Still learning on Railway. Thanks for the correction.

Ok but I saw the answer may take up to 48hours ! And it's a major issue. I will do that anyway.

I filled the private submission form but there a checkbox saying the conversation may become public ! I really don't want to...

Have you try change DNS like I said earlier? Let me know if it's work or not

I didn't have the issue on my machine, so I couldn't reproduce it. And it was easy to get users today... anyway the problem seems to have vanished. FYI we had at least 4 users impacted (home office, so not on the same building and with different ISPs).

You can ask them to try change DNS if this issue is happening again

If it works, You can by cheap domain and proxy it through Cloudflare (as I recommend earlier)

I’m currently having similar issues myself accessing the Railway domain at all. My API was deployed behind a non-custom domain and AT&T is unable to reach it for some reason. Still actively troubleshooting but sounds somewhat related.