Working with Heroku?

Heroku app crashing or bill climbing? We fix it.

Many of the Heroku apps that reach us were set up years ago by a developer who has since moved on, and the bill has crept past what the app earns. We stabilise what is running, cut what is wasted, and move it off cleanly if that is the right call.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

heroku logs --tail

still failing

What it is

Heroku at a glance

Heroku is one of the original platform-as-a-service products, where you push code with git and the platform builds, runs and keeps it serving. It was first released in 2007 by James Lindenbaum, Adam Wiggins and Orion Henry, and has been owned by Salesforce since 2010.

Heroku introduced the Procfile, git-push deploys, buildpacks, dynos and the Twelve-Factor App principles that most modern hosting still borrows from. The free tier ended in November 2022 and many teams have been moving off since, though Heroku Postgres remains a well-regarded managed database on its own.

Type
Platform-as-a-service
First released
2007
License
Proprietary SaaS
Maintained by
Heroku, Inc., a Salesforce company
Typically used for
Rails, Node, Django, Flask, Go apps; SaaS back ends; internal tools

The honest version

Heroku pros and cons

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

Where Heroku is strong

  • Push to deployThe git-push deploy pattern that made shipping feel easy started here, and it is still the quickest way to get a small app on the internet.
  • Good for small appsExcellent when you have one or two services and do not want to think about servers, load balancers or build pipelines at all.
  • Mature buildpacksNode, Ruby, Python, PHP, Java and Go all have well-worn buildpacks that handle dependencies and runtimes for you.
  • Rich add-on marketplaceDatabases, caches, mail, logging and monitoring attach in one command, with billing rolled into the main Heroku invoice.
  • Heroku PostgresThe managed Postgres offering is solid in its own right, with backups, read replicas and a usable dashboard that teams like.
  • Release phase and pipelinesRelease phase runs migrations before new code goes live, and pipelines promote the same build from staging to production.
  • Self-healing dynosDynos restart on their own when they crash, which hides a surprising amount of small trouble without anyone needing to page.

Where Heroku bites

  • Pricing climbed and free tier retiredThe free tier ended on November 28, 2022, and the paid tiers have been raised more than once since. Hobby projects that used to cost nothing now cost real money.
  • Tier model gets complicatedEco, Basic, Standard and Performance dynos each have different limits, and the cheapest ones sleep or share CPU in ways that surprise people.
  • Buildpacks go staleOld Node, Ruby and Python runtimes get dropped on a schedule, so an app left alone for two years often will not build at all until it is upgraded.
  • Thirty-second request timeoutThe 30-second request timeout is hard and cannot be raised. Long reports, big uploads or slow external calls have to be moved to a worker or they fail with H12.
  • Add-on bills on top of dynosPostgres, Redis, logging and monitoring each have their own paid tiers, and the combined invoice is often larger than the dyno line itself.
  • Private Spaces is enterprise-onlyRunning inside a private network, with static outbound IPs, requires the enterprise Private Spaces product and a much larger commitment.
  • Limited regionsOnly a handful of regions are offered, so apps serving users far from them pick up latency that other platforms would avoid.
  • Migration is a projectMoving off Heroku is a project in itself, because the convenience it provided has to be rebuilt somewhere else before the switch.

Sound familiar?

Common Heroku problems we fix

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

H10 - App crashed

The app keeps crashing on boot

Usually a missing config var, a buildpack that no longer supports the runtime version, or a release-phase migration that fails and takes the deploy with it.

Boots clean on every push

H12 - Request timeout

Requests time out at thirty seconds

Slow reports, large exports or a sluggish third-party API that cannot finish inside the hard platform limit.

Long work moved to a worker

R14 memory quota exceeded

The dyno runs out of memory

A memory leak, an oversized worker pool, or a dyno size that was fine for the old traffic and is not fine for today.

Memory profile, right-sized dyno

monthly invoice up 240%

The bill is much higher than expected

A pile of add-ons on paid tiers, dynos scaled up during a spike and never scaled back, and Heroku Postgres on a plan two sizes too big.

Bill cut without changing the app

previous developer unreachable

Nobody can log in to the account

The account is in a former contractor's name, the GitHub integration is tied to their user, and the credit card belongs to someone who has left.

Ownership moved to your team

no migration plan on file

Ready to move off Heroku

The convenience of git push is worth money, but the bill has outgrown the value and the team wants to run on AWS, Render, Fly.io or DigitalOcean instead.

Migrated with no lost data

Node version 12 is not supported

The buildpack rejects the app

The runtime pinned in the project was dropped by Heroku, and the latest buildpack refuses to compile it until the version is raised.

Upgraded and building green

last verified backup: 94 days ago

Heroku Postgres is not actually backed up off-platform

The built-in captures are fine until the account itself is the problem, and no copy exists anywhere else.

Off-platform backups, tested

What we do

Heroku help, from rescue to migration

Call us for one crash, or hand us the whole account.

Rescue
Crashing, looping or timing out. We get the app serving requests again and then find the actual cause.
Cost review
Dyno sizing, add-on audit and Postgres plan review, with a before and after on the monthly invoice.
Takeovers
We take over Heroku apps from previous developers, move ownership to your team, and document what is running.
Upgrades
Runtime and buildpack upgrades for apps that have not been touched in years, done without a rewrite where possible.
Migrations
Move off Heroku to AWS, Render, Fly.io or DigitalOcean, with the database and background jobs tested before the cutover.
Managed Tech
Dyno sizing reviewed each quarter, add-on inventory kept to what you actually use, stack upgrades tested on a release pipeline, and someone on call for 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 Heroku alongside something else?

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

Questions

Heroku questions

Something else? Call+1 737 727 7327

How do we migrate our app from Heroku to AWS or DigitalOcean?

We rebuild the pieces Heroku was providing: a place to run the containers, a managed Postgres, a queue if there is one, and a way to deploy on push. We mirror the data, cut over DNS during a short window, and keep Heroku running until the new stack has proven itself.

Why is our Heroku bill much higher than expected?

Usually it is a mix of add-ons on paid tiers that nobody is using, a Postgres plan one or two sizes larger than the data needs, and dynos that were scaled up during a past incident and never scaled back down. An audit normally cuts the invoice without touching the application code.

Can you take over a Heroku app from a previous developer?

Yes, that is a common ask. We move ownership of the Heroku account, GitHub repository and domains to your team, document what the app does, and remove anyone who should no longer have access.

Why does our Heroku app crash with R14 memory or H12 timeout errors?

R14 means the dyno is using more memory than its size allows, and the fix is either a larger dyno, fewer worker processes, or a memory leak that needs tracking down. H12 means a request took longer than thirty seconds, which is a hard platform limit and usually means the slow work has to move into a background job.

Should we stay on Heroku or move to Render, Fly.io or somewhere else?

Stay if the bill is reasonable and the team values not thinking about infrastructure. Move if the add-on costs have outgrown the value, if you need regions Heroku does not offer, or if long-running requests keep hitting the thirty-second limit. We will give a straight answer for your situation before any work begins.

How much does Heroku help cost?

Estimates are free. Tell us what is running and what is going wrong, and you get a price before any work starts.

Free estimate

Heroku 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.