Running on Render?

Render service stuck or bill creeping up? We sort it out.

Render is often where Heroku apps land when the team wanted the same push-to-deploy feel without the retired free tier. When it goes wrong it usually goes wrong around build caches, free-tier spin-ups or a Postgres plan one size too small, and most of the fix is tuning rather than rewriting.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

render-build.log

still broken

What it is

Render at a glance

Render is a managed hosting platform that runs web services, background workers, cron jobs, static sites, and databases from a Git repository. It was founded in 2019 by Anurag Goel, and markets itself as a modern successor to Heroku with a similar developer flow and a more current price sheet.

Under the hood services run as containers on Render's own infrastructure across a handful of regions in the US, Europe and Asia. The platform handles builds, zero-downtime deploys, TLS, preview environments per pull request, and managed Postgres and Redis. Pricing is a mix of per-service instance types and usage, with a free tier that sleeps idle web services.

Type
Managed platform-as-a-service
First released
2019
License
Proprietary SaaS, mixed subscription and usage
Maintained by
Render Services, Inc.
Typically used for
Web apps, APIs, cron jobs, background workers, static sites, Postgres and Redis for small and mid-sized teams

The honest version

Render pros and cons

No technology is all upside. This is where Render earns its place and where it tends to cause trouble.

Where Render is strong

  • Familiar for Heroku refugeesPoint it at a repo, pick a runtime, deploy. Preview environments, workers and cron jobs all map onto patterns that Heroku users already know.
  • Honest pricing on paid tiersInstance sizes are listed in dollars per month rather than opaque dyno units, so cost planning does not need a calculator with four tabs open.
  • Preview environments per pull requestA fresh ephemeral service and database spin up for each PR, which catches migration problems before they land on main.
  • Managed Postgres and RedisDatabases, backups, point-in-time recovery and metrics without having to stand up RDS. Good enough for most mid-sized production workloads.
  • Private services and a VPCInternal services can talk over a private network without being exposed publicly, which removes a surprising amount of friction compared to older PaaS designs.
  • Simple blue-green deploysA new instance has to pass health checks before traffic shifts, so bad releases are caught before they are serving customers.

Where Render bites

  • The free web tier sleepsFree web services spin down after 15 minutes of idle, and cold starts are visibly slow. It is fine for a hobby demo and not for anything a customer pays for.
  • Build limits bite on big monoreposA 15 minute build cap and a memory-constrained build container are tight for TypeScript monorepos, Next.js projects with large asset pipelines, and Rails apps with a full test suite.
  • Postgres starter plans are smallThe entry tier gives modest storage and RAM, so busy apps outgrow it fast and end up migrating to a larger plan or to RDS before they expected to.
  • Fewer regions than the big threeA handful of regions in the US, Europe and Asia. Users far from them see latency that AWS, GCP or Cloudflare could avoid.
  • Bandwidth overage is easy to missIncluded bandwidth is generous on larger plans and modest on smaller ones, and a chatty API serving large payloads can push past the included allowance without a loud alert.
  • Observability is bring your ownBasic logs and metrics ship with the platform, but anything richer means wiring Sentry, Logtail or Datadog, which the dashboard will not do for you.

Sound familiar?

Common Render problems we fix

The Render problems that bring people to us, and what you have once each one is fixed.

No open ports detected

The service never comes up

The app is listening on the wrong port, usually 3000 locally while Render is probing the PORT environment variable it injects at runtime. Deploys hang and finally time out.

Listening on $PORT, service live

service failed to become live within 10 minutes

Deploys stall on health checks

A health endpoint that depends on a migration that has not finished, or a cold start longer than the health check gives it, so the new instance is torn down before it is ready.

Health checks match reality

Build exceeded 15 minute limit

Builds run out of time

A Next.js or Nuxt project with a large asset pipeline, a monorepo rebuilding everything on every push, or a cache that keeps invalidating. We tighten the cache strategy and split the build.

Builds finish inside the window

Render Postgres: storage 95% full

The database is about to fill up

A starter plan that made sense a year ago and is now out of headroom, with backups eating into the same disk and no alert set up.

Upgraded with a clean cutover

bandwidth: 412 GB

The bandwidth line jumped

An API returning full response bodies where it only needs to return a count, uncompressed JSON, or an endpoint that quietly serves a 20 MB file to every opened dashboard.

Payloads trimmed, cost back down

cron job silently stopped

A scheduled job stopped running

A cron that was set to a timezone the app no longer uses, or a job tied to a free-tier worker that was suspended for inactivity.

Cron running on the right clock

Free plan web service spun down

The app disappears on weekends

The free tier spins idle services down, so Monday morning starts with a cold boot on the first customer request, and the first user of the day pays for it.

Paid plan sized for real traffic

no owner on the Render team

Account inherited from a vanished developer

The GitHub integration is tied to a user who has left, the billing card belongs to someone else, and the environment variables were set from a laptop the company does not own.

Ownership moved, secrets rotated

What we do

Render help, from rescue to upkeep

Pick up the phone for one failing deploy, or let us take the whole workspace.

Rescue
Deploys stuck, health checks failing or a database close to full. We get traffic flowing again and then explain how it happened.
Heroku and Fly.io migrations
Moving onto Render from another PaaS with the Postgres mirrored, preview environments working, and DNS switched on a calm night.
Plan review
Service sizes, Postgres tier and bandwidth usage reviewed against real traffic, with a number on the invoice before and after.
Build pipeline
Faster builds through cache hygiene, monorepo splits, and preview environments that boot quickly enough for reviewers to actually use them.
Databases
Postgres upgrades, backup strategy, read replicas and the occasional migration off Render Postgres to RDS when the workload has outgrown the plan.
Managed Tech
An ongoing eye on instance utilisation, Postgres headroom and bandwidth trends, with somebody to ring on the deploys that misbehave.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Render alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Render questions

Something else? Call+1 737 727 7327

Why does our Render service never become live?

Nine times out of ten the application is listening on a hardcoded port like 3000 while Render is probing the port it exposes through the PORT environment variable. Change the server to read process.env.PORT, or whatever your framework's equivalent is, and the health check will pass on the next deploy.

How do we move from Heroku to Render?

We lift the service, workers and cron jobs across first, mirror Postgres with pg_dump and logical replication where it helps, run both stacks side by side for a short while, and swing DNS over once the new side has processed real traffic. Buildpacks are replaced with a Render build command and a Dockerfile only where it earns its keep.

Why do builds keep timing out at 15 minutes?

Usually a cold cache rebuilding everything on every push, a monorepo without a workspace filter, or a build container running close to its memory limit and swapping. We split the build, pin dependencies, and warm the cache so most deploys finish in a couple of minutes.

Our Render Postgres is 95 percent full, what do we do?

Move to a larger plan during a quiet window with a tested restore. The Render upgrade path keeps the same hostname and credentials, so application changes are usually zero. If the workload has outgrown Render Postgres entirely, RDS or a self-managed cluster becomes the honest answer and we plan that migration separately.

Should we pick Render, Fly.io or stay on Heroku?

Render fits teams who want Heroku's workflow without the pricing climb, Fly fits teams who care about running close to users in several regions, and staying on Heroku is reasonable if the bill is already acceptable and nobody wants to touch it. We will give you a direct answer for your workload rather than a comparison chart.

How much does Render help cost?

The quote is free of charge. Share the service name and what it is doing, and we hand back a plan with a number attached before we ask for access.

Free estimate

Render app giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.