Where should issues be reported?
I reported several issues in this discord group, but it seems that their communication went stale.
Where's the best place to report issues and follow up on what happens with them?
Should I do it on https://railway.canny.io/bug-reports
Or in github at:
https://github.com/railwayapp/cli/issues
GitHub
Issues · railwayapp/cli
Railway CLI. Contribute to railwayapp/cli development by creating an account on GitHub.
46 Replies
Project ID:
N/AN/A
We’re currently working on a better way to track these
Just got like, a LOT of users and working on scaling support/triage/etc
What issues have you seen?
For now Discord is definitely the place but response times will absolutely vary/be slower over the next week or two as we build out the system to scale this out
I'm talking about the thread:
https://discord.com/channels/713503345364697088/1039113839444242442
Where issues with generation of domains that don't always work are detailed
https://discord.com/channels/713503345364697088/1039113839444242442/1039130282235801611
Oh you’re just gonna have to create the domain there
Procfile is something we support for backwards compatibility reasons
I didn't ask for procfile - it was a suggestion by your develper 🙂
Follow this post forward:
https://discord.com/channels/713503345364697088/1043798944628613192/1043799519281823754
The suggestion is to detect there’s a web server and automatically add a domain correct?
Link doesn’t seem to work?
The issues started with a set of steps that after you follow them and setup a custom domain myself - the link doesn't work (you are welcome to follow these steps and see that they still don't work)
Then one of your conductors mentioned that the domain should be generated automatically.
Then one of your developers instructed me to create a procfile.
Domains haven’t been auto created for a while now so I think that’s stale info sadly
If you’d like to create a ticket at https://feedback.railway.app/feature-requests?sort=trending we can track it
Actually no - I don't mind generating the domain.
My problem is that in some cases (like the one I've reported) even after I generate a domain - it doesn't work
Interesting. And you’re listening on process.env.PORT or setting the PORT variable in the dashboard?
I'm listening on process.env.PORT.
You can see my original report here (in this specific message link):
https://discord.com/channels/713503345364697088/1043798944628613192/1043799519281823754
Ah I think I see your issue. It’s not consistent. Sorry just reading through it
That also includes a demo repo I've created for you guys
I think there’s a race with the initial build where if you add it immediately after the builds already been kicked off
Seen this 1-2x before. We’re working on making that better ATM by revamping some of the edge proxy/build workflow.
Could be - although the thread started with a case in which I've waited for the build to conclude before generating dompain
Hmmm
Gotchya. Not ideal. Is see the Template was already mentioned.
I’d create a ticket for this under “bug” and we can surface it to make sure it works with the new networking stuff we’re rolling out
My actual use case are live demos that I create a fullstack app and deploy it.
So far I've used Heroku, but I really want to use railway.
But in a 30min live demo, I can't afford to use something that is not consistent 🙂
Where would you like me to create the ticket?
Totally reasonable!
And here's the second one:
https://feedback.railway.app/bug-reports/p/databaseurl-missing-if-database-is-deployed-after-app
Awesome thank you! Will include these in our triage call Monday 🙂
Can’t promise any resolution time but letting you know it’s noted, and we’ll be able to follow up via the ticket once we have movement on em!
Thanks
Not a conductor; employee 🙂
I have the call in 30, will be discussing this with the team.
I was referring to this message
https://discord.com/channels/713503345364697088/1039113839444242442/1039132275524247572
🙂
lololol
not for existing templates
meaning: we won't autogenerate the domain for your demos, and unfortunately we have too much work in other places to ship something of the sort anytime soon 😢
That's not a problem - I can click generate domain, my issues was that after I manually generate domain it doesn't always work 🙂
D'oh
How long are you usually seeing it until it finally registers?
I tried deploying through here and still didn't get the domain - also when I deploy from here there is a problem, since it first deploys my code - and only then I can add a database - but there's another issue where if I deploy the database it doesn't redeploy the app - so the app crashes for lack of database 🙂
Yea, its coming back to me
If you click generate domain before the build completes you can wait for hours and it still won't come
And in the problem I had when I started this thread, I waited for 24 hours before trying multiple redeploys until it eventually worked
When you apply a custom domain, how long do you notice that to be registered?
Just trying to collect more data for the team here.
When I click on "generate domain" I get it within a few seconds - then I open the link and wait.
1. If I clicked generate domain before the build was completed (and rested) I can wait for as long as I wont - it'll not work
2. If I wait for the build to rest and then click generate domain - in most cases after I'll click on the link it'll wait for a minute or two and then will open - but not always
Gotcha, 1. should absolutely not be happening.
2, that seems in line with our proxy behavior, not ideal, but we have a long term fix for this.
I am going to escalate 1 internally with our Network Engineer and see what we can do to get to the bottom of this.
Thanks.
About 2, It's a bit stressful in a live demo not to have a consistent behavior where it'll sometimes work and sometimes not - is there a sliver bullet I can use once it doesn't load?
So far I was stumbling with editing the domain name randomly, or multiple redeploys etc...
Here's a video of a deploy I've done in a recent meetup:
https://discord.com/channels/713503345364697088/1039113839444242442/1039122337922035743
Redeploys infact make the domain issue worse because we need to reattach the domain for every new deploy on the proxy, so for demoability I wouldn't recommend that.
I totally get the feeling of having things work on stage, that was my job for a bit.
I would just wait 3-4 minutes. Changing things while we are attaching your domain makes things thrash more.
Which seems to explain the unpredicatable behavior.
Last time that it didn't work - I waited for 24 hours
before redeploying it
Yes, ofc that shouldn't happen, but changing properties definitely doesn't help.
Lemme discuss this with the infra team, I am sure we can get you to a good resolution.
Sounds great - thanks, I appreciate your responsiveness
But the short term work around would be using a custom domain, that would short circuit our proxy.
More crisp of a demo for sure.
You sure on this? I think it uses the same upserting flow :/
This is my 90 percent guess, the problem is the envoy table is my hunch.
ofc Charith might know more
I'm not sure I follow. I thought that "custom domain" is used to bind to my own domain - not for subdomains of railway.app (forgive me if I cross terms)
In my live demo I write a fullstack app and deploy it - I need an ad hoc url to use in my presentation 🙂