The live site or API is down, or so degraded that customers cannot use it.
One free session per client: the first 4 hours with a specialist, plus unlimited AI as a tool.
A real senior specialist leads. Name on first reply. AI is a tool; a human checks the result.
01
What this problem looks like, how you can tell, and what is at stake if it stays.
What a production outage looks like, why every minute feels public, and why a calm specialist is useful in the middle of the noise.
Health checks fail. The load balancer has nothing healthy behind it. A deploy went wrong. A certificate, a disk, a database, a DNS change, or a dependency took the service with it. Status pages are already lighting up.
Every minute feels public. You are being asked for an ETA you do not have. The team is in a call and still arguing about the first step.
Uptime is a promise you already made to your users, even if it was never written down. When production is down, that promise is already broken. The work is to restore it honestly.
You want it back, with a cause, not a reboot that will fail again in an hour.
When production is down, the live site or API is offline or so degraded that customers cannot use it. You want it back with a cause, not a reboot that will fail again in an hour.
A deploy, a certificate, a disk, a database, a DNS change, or a dependency can take the service with it. Status pages light up. The team is already on a call and still arguing about the first step.
Weekend work is a separate product; any surcharge is stated up front, before a call. We do not take every support request. We take a hard down system that needs an engineer.
These are signals that you are in an outage, even if one region or one path is still up.
Instances fail, or they start and die in a loop.
A marketing homepage may still be on a CDN while the app is dead.
5xx, timeouts, or connection refused, across more than one user.
Some nodes moved, or a migration is mid-way, and nobody wants to touch it.
DNS, a database, a queue, or a partner API is the real outage.
The reboot was not a cause, only a pause.
Before — if you leave it
While it is down you lose sales, trust, and time from people who will remember the silence more than the later explanation.
After — if we close it
Service is back, the immediate cause is known, and the unsafe workaround — if one was needed — is written down so it can be replaced.
Recognize your situation? Message us, we will sort it out.
02
Where this usually comes from, why it shows up in fast or AI-assisted work, and how to check it yourself.
Most outages have a last change, a saturated resource, or a dependency. Finding which one is the work.
A flag, a secret, a certificate, or a container image that only fails at live scale.
Connections, locks, a full volume, or a migration that will not finish.
A record, a certificate, or a load-balancer rule that no longer points at healthy origins.
A cloud limit, a payment of the account, or a third-party outage with no timeout.
An attack, a traffic spike, or a process that ran away after a deploy.
A generated pipeline or a 'works on my machine' container will often meet production for the first time on a launch day. Fast startups also skip runbooks, so the first outage is also the first rehearsal.
A Human First incident still starts with the last change and the logs, not with a hopeful restart script written by a model in the middle of the call.
Capture evidence as you go. A reboot can help and also wipe the picture you needed.
Which URLs, which regions, which users, and whether it is total or degraded.
Deploy, DNS, certificate, cloud change, or a job. Write down the time.
One of those is often already telling the truth.
A known last good artefact is a gift, if data allows it.
Defacement, ransom, or unexpected admin. That changes the order of work. See also the hacked-website page.
Recognize your situation? Message us, we will sort it out.
03
A senior specialist leads. AI is a tool. A human checks the result.
Outages are chaotic. A calm specialist who has seen this shape of failure is useful precisely because you are in the middle of it.
DevHero is Human First. You write, you see a specialist's name, and that person can be alerted and called. AI may help us scan logs. A human leads the recovery and checks the fix. We do not take every support request. We take a hard down system that needs an engineer, not a help-centre script.
You write, you see a name, and that person can be alerted and called. AI may help us scan logs. A human leads the recovery and checks the fix.
How a typical weekday case moves
Write on WhatsApp or Telegram now. If you need to be first in the queue, ask about the urgent move to position 1. Weekend work is a separate product, with any surcharge stated up front.
Times, URLs, last deploy, and whether you rolled back.
Within a minute you should see their name. They can be called if they do not respond.
A longer reliability programme is a complex project.
We join the incident, restore a safe path to live, then tell you what still needs a lasting fix. A human checks the recovery.
We join the incident, look at the last change, the dependencies, the logs and the health of data, then restore a safe path to live. After it is up, we tell you what still needs a lasting fix. A longer reliability programme is a complex project. Weekend work is a separate product; any surcharge is stated up front, before a call.
Last change, scope, and what has already been tried.
Rollback, failover, or a containment that brings the user job back, if that is safe.
Logs, health, disk, dependencies — without destroying the only copy of the trail if we can avoid it.
Some causes need a follow-up. We will not pretend a reboot is a root-cause analysis.
So it can be replaced.
The first aim is to restore a safe path. The 24-hour aim, on a typical weekday case in scope, is a cause and a checked recovery.
A specialist on the incident, named, once you are in the session.
Service restored, or a clear containment, if that is possible with the access we have.
A cause, or a narrowed set of causes, in plain language.
A note of any workaround that must not stay.
A human-checked recovery, and a follow-up if reliability work remains.
Speed matters. A few facts prevent a blind restart. Do not paste live passwords into the chat.
Plus the time zone.
Errors, blank pages, login failures, payments failing.
Deploy, DNS, certificate, or cloud event.
Restart, rollback, scale-up, or a failed failover.
An agreed channel and authorised access.
First 4 hours with a specialist + unlimited AI: free, once per client. Further work is paid, on request. A human leads and checks the result.
Free (first session)
Larger work uses a different rate, on request. We say so before we continue.
On request
The base plan is weekdays only. Weekend work is a separate product. Any surcharge is stated up front, before any call, and depends on complexity. Amount: on request.
On request
The first session is free: the first 4 hours with a specialist, plus unlimited AI, once per client. Work beyond that, the complex-project rate, and the weekend rate are paid, on request.
One free session per client: the first 4 hours with a specialist, plus unlimited AI as a tool.
Recognize your situation? Message us, we will sort it out.
04
What not to do next, the stacks we see in this kind of work, and how to keep the fix in place.
Motion is not the same as recovery. Some moves make the next hour harder.
You may get a brief recovery and lose the cause.
You can lose the last good artefact.
Restore first, then write.
You may be wrong, and you may reveal a path you have not closed if this is also a security event.
If the site was defaced or data is leaking, see the hacked-website and security pages and say so early.
Tell us the stack and the host. If we should not take it, we will say so.
AWS, GCP, Azure, and simpler hosts.
Postgres, MySQL, Redis, and job workers.
The edge that can look like an application failure.
A CI/CD change that became an outage. See also failing deploys.
A recovery is incomplete if the same move would take you down tomorrow.
Cause, workaround, and the lasting fix.
A rollback you have never tried is a hope.
Not only on a static page.
Boring checks.
When you can.
A name and a path, not a group chat with fifty people.
A pipeline you do not understand will fail in public.
Recognize your situation? Message us, we will sort it out.
05
Straight answers about this problem, the session, and what happens next.
If you have a known last good release and a safe rollback, say what you already tried.
Say that. We will treat it with that order of work. See also the pages on a hacked website and on security holes.
Yes. Ask about the urgent move to position 1. The start price is 99 USD while the queue is empty, and it can be revised later.
Write on WhatsApp or Telegram now. No form, no web chat.
No. The base plan is weekdays. Weekend work is a separate product, and any surcharge is stated up front, before a call.
No. We will talk about this incident and what we changed.
If you want a reliability follow-up, we offer the matching rate and we sign for that work.
Not if the user job is down. Health checks can be looking at the wrong thing.
No. We take a hard down system that needs an engineer, not a help-centre script.
Yes. Each client gets one free session: the first 4 hours with a specialist, plus unlimited AI as a tool. The 24-hour line is how quickly we aim to start and close a typical weekday case in scope, not 24 hours of unpaid work. Further work is paid, on request.
If production is already down, write on WhatsApp or Telegram now. You will talk to a person.
One free session per client: the first 4 hours with a specialist, plus unlimited AI as a tool.