/bells chime, doves are released, balcony door flings open, crowd looks up expectantly 

Hellooooooooo! It’s your friendly neighborhood changelog author here!

That’s right folks, today’s changelog is all about speed and there’s nothing more 

/crowd empties out, tumbleweed rolls through the square, pigeon lands on changelog author

No but seriously folks, we do have a great one for you today.

We’ve solved some major dashboard latency issues, we’ve increased single-threaded workload speeds on metal by 30%, we’ve shipped a Mongo optimization that reduces costs by more than 50% … and, well, you’ll have to read on below to see everything we did this week.

Before we get started though, we wanted to remind you that 

 continues onward. Most of our team is working on the migration right now so we want to know ANY issue you run into. 

Seriously, if you have an issue migrating over to Railway Metal, make a ticket in 

. We are prioritizing all metal reports until they go to zero. 

 for everything you need to know about swapping over to Railway Metal. 

We’ve eliminated most dashboard latency spikes

White smoke was coming from our ears a bit over the past couple weeks as we navigated some tricky latency issues with the Railway dashboard.

But the smoke settled and we’re happy to kick off a new era of speedy dashboard performance with a fantastic new dashboard performance experience. 

If you’re curious, we solved a number of issues. One issue was related to our own internal code deployments. We deploy a lot of new code to Railway — many times a day, every day — and shipping those updates to many thousands of user sessions started causing a thundering herd problem in the canvas. 

Another problem we solved involved redundant queries to some external service providers and internal APIs. While yet another one involved the way we were handling websocket connections for stale sessions.

… all in all, we’ve instituted a half-dozen or so optimizations that have resulted in a 

 the latency blips that were impacting the dashboard over the past couple of weeks.

If you still run into any latency issues, we want to know about them. 

 50% cheaper on Railway Metal (compared to non-metal instances) due to some new optimizations we made.

Well, as we’ve been saying for a while, Railway Metal allows us to deliver service that will ultimately be better, faster, cheaper, more configurable, and more reliable than our legacy “

This week, Mongo is an example of that. We found when running on metal that a certain default Mongo flag was improperly set. By changing the flag for default Mongo installs, we save you money. 

Bounties are now available to question-answerers in Central Station

If you’ve been a Railway passenger for a while, you should be familiar with 

It’s our one-stop shop for all your help and support needs on Railway … and since the day we decided to 

 entirely, we’ve gone from solving dozens of tickets a day to many hundreds of tickets a day. 

As a result, we need to call in air support. 

From today, YOU can now answer public support tickets on Railway and we’ll pay you in credits to do so.

Select your workspace to receive reward credits

If your answer is accepted, you’ll get 

That $10-per-thread bounty is HUGE and it probably won’t last. But it’ll last at least this month as we seed the program before we start making it more reasonable. 

Sound like an interesting way to help people out in the Railway community and get rewarded for doing so? We think so too.

. This is useful if you need to run long-running tasks inside the deployment and want to make sure the command isn’t interrupted

 so be sure to check that out if you haven’t seen it in a little while

https://res.cloudinary.com/railway/image/upload/v1746839677/2025-05-09_unvdhl.png

Dashboard Latency No Longer, 50% Cheaper MongoDB, Bounties for Answering Questions

Changelog #0237