Server error but no logs found
Service ID: 50bac420-0946-41bc-9b51-f3bda884becd
Solution:
Jump to solution
To close this topic and contribute to future similar cases:
The real problem was that the site was suffering a DDoS attack, which caused the process to simply freeze, this was only noticed today at 10 am as the resources on Vercel's Hobby plan had been completely exhausted. The attack in question has already been resolved after increasing the security level at the DNS provider (Cloudflare) and configuring the rules for the API.
After the end of my vacation I will possibly return the service to Railway, thank you for your availability....
82 Replies
Project ID:
50bac420-0946-41bc-9b51-f3bda884becd* App works perfectly in localhost, no changes are maded in last 10 days, suddenly started
what kind of server error?
zero logs of any kind?
0
nothing
that's definitely not zero
yep, it's zero, the first line occours since 10/2022
The timestamps say jan 2nd… that’s today
I’m a bit confused by what you mean by zero logs when there’s clearly logs
error log
says server error, but in output no errors are showed
doesn't always mean there are going to be error logs
what's your start script in your package.json
you are listening on localhost, listen on 0.0.0.0 instead
fairly certain next does listen on 0.0.0.0 anyway but just says localhost so you can actually visit the logged link. but what I don't think next does is listen on PORT, and for that we need to wait until my question is answered
I love documentation
Has the railway made any infrastructure changes in the last 24 hours? because it doesn't make any sense that everything has stopped working.
what is your start script?
no, they haven’t
Apparently my build configurations were deleted from the dashboard...
I'm still checking
should be in your package.json
Nothing has changed, even when returning with my personalized configuration I still receive "Application failed to respond"
All scripts are Next.js defaults (next build/start)
Caillou, please send your start script
Everything else is done by Nextjs itself
update the start script to what is in the image adam sent please
^
nothing changed...
do you have any build or start commands set in the service settings?
only build
do you have a Procfile?
no
build logs please https://bookmarklets.up.railway.app/log-downloader/
you are overwriting your start script, you have a start command set elsewhere
Yes, because you asked earlier
Regardless of the configuration used, it doesn't work, this is not within my power, the service simply stopped working without any changes to the code/dashboard.
show me your scripts please
.
These are the default scripts, all changes were made trying to solve this problem.
please follow my instructions
We asked for you to change your start script in your package.json to --port ${PORT}
this do not work
not to -p 3000
Nextjs startup parameters are different from node, -p is equivalent to Node's -port
this is not my first rodeo
The instructions I sent are specifically for next
Please follow them
and please remove any start command you may have set anywhere other than in your package.json
The build process is insanely slow, it took almost 15 minutes and we didn't even get to the build of the project itself
It seems to me like it's in a loop, every 5 minutes it prints:
on stdout
your issue is not related to theirs
please do no hijack others threads that are unrelated
moving on, show me a screenshot of the repeated logs please
14 minutes and the build process wasn't even halfway through
odd but looks harmless
This normally takes a maximum of 5 minutes
@matt - possible builder slowdown
build finished in 20 minutes, waiting for deploy ending...
Even using the specified parameter the problem persists
show me your scripts please
that is not the correct start script
??
I don't know how you think you have the same start script
My friend, the service worked until 18:05 (UTC-3), at 18:06 it stops working without any changes, would this really be my configuration problem?
yes 100%
please use the correct start script as shown in the docs and the screenshot
I'll arrange a migration to Vercel, I've already had a huge loss trying to resolve this, thanks anyway.
it's a simple configuration issue
Agreed, we’ve told you to change your script to pull from the PORT environment variable. NOT set the port to 3000
this is an extremely simple change
I'm sorry you've had difficulties but we have only tried to help you and we have been meet with pushback
I tried the last time and the problem still persists, even though I set the dynamic port as requested.
show me your scripts now please?
Aqui está minha ID TeamViewer e senha de conexão.
Por favor, use esses dados para iniciar uma sessão de suporte remoto comigo.
ID TeamViewer: 1 261 830 xxx
i dont have teamviewer downloaded right now, give me time to download it and id be happy to help you that way
looks like there was a problem with a single builder ~40 minutes ago but that was resolved and everything looks good rn
perfect, thanks for checking
update after teamviewer, there is a bigger issue at play with their code, it uses far too much cpu and mem and is softlocking itself
great work here, Brody
no not great, i did not find a fix
they are going to try deploying to vercel, vercel will pull some magic strings and their code will likely work just fine over there
I was transferring the DNS settings and noticed that Railway had started working again, but after 1~2 minutes it stopped again. Quite strange
Ah, it just started working again, it looks like some kind of wobble
while I agree this is super strange, from looking at your service metrics the issue is something along the lines of your app is soft locking itself as opposed to an issue with railway, and again I wish I knew the issue so I could help you more here
Solution
To close this topic and contribute to future similar cases:
The real problem was that the site was suffering a DDoS attack, which caused the process to simply freeze, this was only noticed today at 10 am as the resources on Vercel's Hobby plan had been completely exhausted. The attack in question has already been resolved after increasing the security level at the DNS provider (Cloudflare) and configuring the rules for the API.
After the end of my vacation I will possibly return the service to Railway, thank you for your availability.
oh wow, so I was only half right. your app was indeed soft locking, but it wasn't from any bad code. I'm deeply sorry for making that particular assumption
No problem, it's part of the job, it's a really difficult scenario to make any assumptions about.
well I'm glad you where about to figure it out!!
can you mark solution? the option do not show for my
yeah only mods can mark solutions, had too many users marking the wrong message as the solution.