Skip to content
Very long build queues
mbirkegaard
OP

a month ago

We're currently seeing builds queued in a preview env queued for 30+ minutes and no mention of slow builds on the status page. Is anything going on? If not and there's just a lot to go through, consider this a feature request for something that estimates the number of builds in front of mine.

Solved

15 Replies

Status changed to Awaiting Railway Response Railway • about 1 month ago


Railway
BOT

a month ago

Your builds were queued because the workspace had reached its Pro limit of 10 concurrent builds. Around 20 builds were triggered at nearly the same time, so the extras had to wait for a free slot. That queue has now cleared and nothing in the workspace is waiting for a slot.

If every service in an environment rebuilds on each push, setting watch paths on each service means only the services whose files changed get rebuilt. That keeps fewer builds competing for slots.

Thanks for the suggestion about showing how many builds are ahead of yours.


Status changed to Awaiting User Response Railway • about 1 month ago


Status changed to Solved mbirkegaard • about 1 month ago


mbirkegaard
OP

a month ago

We're still seeing huge queues and builds being slow, so nothing is clearing. That there's a queue that's fine, but builds getting stuck for 1 hour+ and then blocking everything else seems not intended?


Status changed to Awaiting Railway Response Railway • about 1 month ago


mbirkegaard
OP

a month ago

Is there some way to autocancel queued builds so only the latest remains for any service?


Railway
BOT

a month ago

There's no setting that automatically cancels older queued deploys so only the latest one per service is left. Thanks for the suggestion.

The queue comes from your workspace's limit of 10 concurrent builds on Pro. Right now all 10 slots are building and 2 deploys are waiting. Most of that is a single push at 12:05 UTC that started a rebuild of about a dozen services in one preview environment at the same moment. The build that had been running for over an hour earlier today has now finished.

To get fewer builds competing for slots, set watch paths on each service so a push only rebuilds the services whose files changed. Also delete preview environments you no longer need, since each one rebuilds its services on every push. Smaller images help with the slow builds, because every build has to package and upload its image before the slot frees up.


Status changed to Awaiting User Response Railway • about 1 month ago


mbirkegaard
OP

a month ago

I miss when Railway used to be easy, now I have to think about devops things that I didn't want to have to think about


Status changed to Awaiting Railway Response Railway • about 1 month ago


mbirkegaard
OP

a month ago

Builds are triggering multiple queued commits, so our limit gets eaten up super quickly. What is going on?


mbirkegaard
OP

a month ago

This is a freshly opened preview env. Other envs are seeing 3 x builds of the exact same commit being triggered at exactly the same time

Screenshot 2026-10-14 at 15.08.43.png


mbirkegaard
OP

a month ago

We closed down all but our staging and prod envs and across the two there are 24 queued deploys and 3 building deploys. Something is off and it's preventing us from doing our work.

Edit: Got this wrong. There are queued builds in another env, so we're at the limit now. All builds are progressing super slowly


a month ago

Not having the queued issue, but build time is slooooooow … already wondered whether the status page tells anything.


mbirkegaard
OP

a month ago

Yeah, we're seeing very slow builds too, which is worsening the problem!


a month ago

usually it's done after max 4 minutes

image.png


mbirkegaard
OP

a month ago

Indeed

Screenshot 2026-10-14 at 15.32.58.png


a month ago

Your 10 build slots are shared across every project in the workspace, so the 3 building deploys you see in staging and prod are only part of it. Right now all 10 slots hold builds from a single push at 12:35 UTC, and 31 deploys are waiting behind them. That includes a second round of the same services from a push at 12:43 UTC and more from 13:00 UTC. An older queued deploy still builds before a newer one for the same service, so each extra push adds builds rather than replacing them.

In the queue we can see, each push creates one deploy per service per environment. Two queued entries with the same commit message are usually the same commit landing in two environments, or two pushes queued minutes apart.

The slots free up slowly because each build runs over 20 minutes. In the build we looked at, installing dependencies for the whole repository took about 4 minutes, and copying node_modules into the image took about another 4, while the service's own compile step took about 2 seconds. Installing only the dependencies each service needs shortens both steps, and watch paths stop a push from rebuilding services whose files didn't change.

For anyone else seeing slow builds on their own project, this thread covers one specific workspace's setup, so a separate thread with your project link is the best way to get your builds looked at.


Status changed to Awaiting User Response Railway • about 1 month ago


mbirkegaard
OP

a month ago

Yeah, I realised it was across all projects https://station.railway.com/questions/very-long-build-queues-34c32ce8#zrsv

I still saw multiple builds also in fresh preview envs https://station.railway.com/questions/very-long-build-queues-34c32ce8#9708

The slots free up slowly because each build runs over 20 minutes. In the build we looked at, installing dependencies for the whole repository took about 4 minutes, and copying node_modules into the image took about another 4, while the service's own compile step took about 2 seconds.

That doesn't quite add up to 20 minutes, but in any case, we'll try to optimise. Still feels slower than usual, but it may be deceptive then.

Do you have some kind of internal UI or diagnostics with which you look at this? Just clicking around the canvas and/or using the CLI does not give me a coherent and neat view of what's happening. Can we natively instrument the builds and deploys?


Status changed to Awaiting Railway Response Railway • about 1 month ago


a month ago

Fair point on the math. That build also copied node_modules into the image a second time (another ~4 minutes), and there was a stretch of about 7.5 minutes between the compile step finishing and the next copy where the build log shows no output, so the log alone doesn't tell us what that time was spent on.

For a single view of what's happening, webhooks are the native option: they're configured per project, fire on every deployment status change across all environments, and the payload includes the service, environment, commit and timestamp. Sending those to Slack or your own endpoint gives you a timeline of queued, building and finished deploys for each service. Delivery is best-effort, so for anything exact you can reconcile against the public API. Per deployment, the deployment view shows each phase with its duration, and the Log Explorer lets you query logs across all your services in one place.

Right now all 10 slots are building and 5 deploys are waiting.


Status changed to Awaiting User Response Railway • about 1 month ago


Railway
BOT

24 days ago

This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!

Status changed to Solved Railway • 24 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...