Such a button either doesn't exist, or the option is buried. The competitors to Railway has this feature. Please add this feature. I don't like having to delete the app and then re-creating it to get a fresh instance. Making the app rebuild on each code change is too heavy and I want to exactly control when the app get's rebuilt.
#Railway app needs a force deploy button like render and digitalocean
71 messages · Page 1 of 1 (latest)
It's already planned. Railway will add a redeploy/rebuild button soon™️.
sweet
If you want to control when it’s rebuilt use the CLI
You should be going off a trigger. A manual redeploy button is a backstop.
What are you trying to do instead?
Believe it or not, i do everything windows and only go to a linux VM as a backup for development
and the CLI doesn't work for windows
CLI definitely works on Windows
What are you doing that needs the application rebuilt?
Off not a change to the application itself
i'm testing a social media app called phpFox and I need to restart the containers from the fresh docker instance every now and then
You should fix your code then
Which will retrigger a build
I’ve never encountered an issue here that wasn’t user error. Answered this a great many times as well
right, that's the problem. I don't want it restarting every time i push an update to a readme file
Then change your watchpath
thanks for the work arounds!
anywhere I can update the redeploy button though? It seems that most of your competitors has it except railway
A rebuild button would be helpful in cases where railway has an outage or an external third party (example github/npm/yarn) goes down.
Just cause your competitors jump off our a bridge doesn’t mean you should
Is this Jake Cooper?
There’s no point in building for exceptions
Yep
ha, i was watching one of your youtube videos right now, what a coincidence
Don’t get me wrong. We’ll add one
I’m gonna be annoyed about I though. I think it’s a massive anti pattern and a symptom of failure
Deleting my code repo or changing whitespace to trigger my build is a worse anti-pattern
The anti pattern is you’re deploying manually
Or haven’t configured your watch path correctly
here's my stages of app development
stage 1: redeploy all the time on each code change because i'm building new stuff
stage 2: ok cool we are getting more mature and the code changes are less drastic, I am still developing, but i only want updates when I trigger them manually
There’s the anti pattern
stage 3: the production phase - Harden
You’re shooting yourself in the face by manually deploying as a rule
I think there are two cases. One where your hacking around making a side project and one where you're deploying a massive app. And railway is focusing on the latter?
Would that be a right assumption?
Each change should be versioned and built
yup
One is a superset of the other
Yee
Totally agreed (this shouldn't be a priority) but it'd just be a lot more convenient if railway had a rebuild button in situations like #1025534649268699157 message
There’s literally zero reason to switch to cutting releases manually
Use a different branch
Yup
I'm just explaining my workflow, I'm sure this is an extremely common usage pattern
i don't see how it's an anti-pattern at all
it works extremely well
Railway is about instilling best practices
You want to hook your customers at the early stage right?
You create a bug in between your deploys
Instead of knowing when that bug is introduced, you now have to bisect every commit
If we have to “hook” our customers we’ve done it wrong
auto-rebuild on code change is also kind of a anti-pattern
It’s either applicable at all points in the journey and people learn how to do development the correct way
Or it’s not and we’ve failed
It’s not it’s industry standard
Anyways, I’ve gotta go to some work. We’ll add it at some point
if it's not the industry standard, why does Render.com and DigitalOcean (and maybe fly.io) all support it?
Fly does not
And it’s cause those other platforms are just Heroku clones, which comes from hobby users asking for things that will 1) break workflows and 2) shoot them in the foot
We’ll add it. It’ll be in the command bar at some point in the future.
But if you’re reaching for it often, we’re either fucking up on our engineering side (because we had an outage) or you’re going to run into an outage/issue in the future that’s going to teach you how to do development correctly
Also, one more question, would you ever consider doing a healthcheck on port 80 if port 8080 isn't being served by the docker app?
No
k
Well, sorry that’s not true. We’ll allow you to do multiple ports in the future
So, you could have it dial url/health and that would cover it
right, right now all the major services except Render.com requires re-mapping the health check port manually
It's a poor experience though on all the platforms because the app just fails. It's a beginner foot gun but maybe not something you are aiming to fix
anyway, thanks for answering my questions
😭
i dont agree with you on this