Railway favicon

Railway outage on 2026-10-20

railway.com
October 2026 20 Major incident

Railway paused new builds for most workspaces for 3h 53m on 20 October 2026. Services already deployed and running kept working, but anything you tried to deploy or build during the window had to be queued and retried after the fact. The fault sat in Railway's build pipeline, not in your code or your running services.

Started

17:03 UTC

Duration

Lasted 3h 53m

Source

IsDown

Next time

Railway is degraded. Is your product working?

An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.

50 monitors for your own websites. No card.

What happened?

Railway's status page opened the incident at 17:03 UTC, reporting an issue affecting builds and announcing that new builds were paused for most workspaces while they investigated. At 17:18 they said they had identified the issue and, rather than narrowing the pause, widened it to all workspaces while they worked on a fix. Railway was explicit throughout that services already deployed and running were not affected, only new build and deploy attempts. At 19:10 the status page confirmed builds were still paused for everyone, with a plan to bring them back in stages rather than all at once. A fix landed by 19:59, and builds resumed first for Enterprise workspaces, then Pro at 20:20, then Hobby at 20:43. Each stage required anyone who had triggered a deploy during the pause to trigger it again, since queued attempts from the outage window were not automatically replayed. One user report in the window, from the United States at 18:33, read: "New deployments blocked. No ETA on fix.." That matches what the status page was saying at the time, no ETA had been given yet. There is no components list published alongside this incident, so Railway did not break out which part of the build system was affected beyond calling it "builds". No post-mortem has been published explaining the underlying cause. The incident closed out at 20:43 with Hobby workspaces back online, and there is nothing further filed after that.

Learning

This outage sat entirely in Railway's build pipeline, the service that turns your pushed code into a running deployment, not in the containers already serving traffic. That is why Railway could say with a straight face that running services were unaffected while simultaneously blocking every new build for almost four hours. If your own uptime checks only ping your live endpoints, they would have stayed green the whole time, because those endpoints genuinely kept working. What they would have missed is that any deploy you tried to push was silently queued and then discarded, needing to be re-triggered once Railway staged builds back in by plan tier. That is a failure mode that does not show up as downtime on your product, it shows up as your team wondering why a fix they shipped two hours ago still isn't live. Watching your own endpoints tells you your product is up. It does not tell you whether your provider's build system is the reason your latest change never reached them. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.

4.7
stars out of 5
284+ reviews on

Start monitoring in 30 seconds.

There's nothing to install. No credit card required. 50 monitors for free.