The Railway team is knee deep in grease to make the platform sing while we have been experiencing a massive influx of users. GA tickets remain available for all.

A theme of this week was finally confronting the parts of the Railway platform that begged for replacement making some very scary noises. 

And… spoiler alert, upgrades have been made. We have reliable crons, snappy builds, faster proxy, and faster… well… everything. 

We will be going deep into what went into these upgrades and the features that hit GA.

e truly encourage every company to reach out to us

 to set up a Slack Connect channel. Throughout this period we have bumped up the amount of engineers on-call in support and infra to respond to issues on or off the platform.

We are truly excited for the future of this platform, and more excited about it’s present- so let’s talk Changelog shall we? 

Railway was like a 2013 Toyota Prius with a burnt out hybrid system. As of today, we think it’s now more like a Lucid Air.

You should notice right away a much snappier dashboard experience with faster builds, faster status response times, faster publishing, and overall a much better experience. 

The Railway Infrastructure team pulled day to night coordinated shifts overhauling our DB, crons, builders, and runtime for most of the fleet. Don’t believe us? Numbers.

The dashboard now delivers sub 100ms response times for p99 of requests.

Domain discovery time improved from 2 minutes to 

Build times have been reduced on avg. by 30 percent.

Publish times halved, in some cases, sometimes non-existent for fast follow builds.

Intermittent workload 503s have been eliminated vs. the legacy system.

A cause of intermittent Dashboard 404s have been eliminated.

Patched our cron scheduling logic to clean up prior deployments that may have prevented upcoming executions from properly running.

Workload performance is improved, your code should just run faster. No code change needed.

All of this thanks to the feedback and testing of V2 from our Priority Boarding users. The Infra team also zapped 35 major bugs as well. 

(Reminder to the Infra team, update your Linear!)

Graph lovers will also rejoice in our DB load charts. Bring your best workloads, we can take em.

Our DataDog Graph of our Dashboard Database performance before and after DB upgrades.

And we still have much more coming down the pipe still. 

Please yell at us, or whisper, we’d argue our hearing is rather great

These upgrades also come with a new dashboard in town — it’s bigger, better, faster, stronger, and now comes with Service Groups included. 

That’s right, you can now group services together to your heart’s delight.

Compose a new group of services by right-clicking the dashboard canvas

In addition, you’ll notice that connection lines are indicated for variable references. This can be toggled on or off in the left sidebar. 

In the future, we’ll be adding lots of additional features to the connective portions of the canvas (think 

), but for now we’re leaving it at variable references.

Let us know what you think of Service Groups by leaving feedback 

The new Observability Dashboard is configurable to your needs

 we launched a new Observability Dashboard into Priority Boarding. 

This week, we’re making that dashboard generally available. 

As we mentioned before, we think you’ll especially enjoy all the configuration options we’re providing for adding widgets (CPU Usage, Memory Usage, Network In/Out, Disk Usage, Logs, and Project Usage).

If you have feedback, we’d love to hear it. Please let us know 

 we told you all about the new Edge Proxy that we (then) rolled out in beta. 

There is a major update, new domains on Trial and Hobby will now use it by default.

As a reminder, the Edge Proxy is responsible for routing all external traffic to your application’s server and domains. 

 and earlier above the fold, there are some immediate benefits to you:

You’ll experience near instant deploy cutovers instead of waiting for a minute until it’s provisioned as currently happens today

You’ll get new detailed error pages with a traceable Request ID that you can forward to us when debugging an issue

And there are some benefits that this will unlock down the line:

We’ll be able to further improve service start time

We’ll be able to provide automatic port detection

We’ll be able to provide network logs and metrics

We’ll be able to add support for advanced routing rules and load balancing strategies

We’ll be able to improve service quality and DDoS attack mitigation options

And just to reiterate, the new Edge Proxy is only enabled by default for 

. Your existing services will continue to use the former edge proxy unless you’ve opted into the new proxy via service-level networking settings.

Let us know if you have any feedback on the new Edge Proxy by dropping us a line over in 

We configured template services to use the 

We improved the in-project view when using mobile

We fixed some little visual nits when using light mode in the dashboard

Help Station: we now have a project dropdown within the Thread modal. No more back and forth with the team asking you which project you are talking about.

Various stability fixes and improvements.

Provisioned additional infrastructure and implemented solutions from the June 11th incident report.

https://res.cloudinary.com/railway/image/upload/v1718354021/2024-0614_rrbmjg.png

Faster Everything, Service Groups GA, Observability Dashboard GA, Edge Proxy GA

Changelog #0190