How to run things on StackHost, in plain steps.
Short guides for the things people actually do here: launch a first site, point a domain, restore a backup, deploy with Git, size a server, read a bill. Every guide stays on the published terms — when something needs your account details, it says so and points to your portal.
26 guidesSupport hours Mon–Fri, 08:00–18:00 EATSupport & SLA
First steps.
What to set up first after you sign up
Create your account and choose a plan. Starter covers two sites and carries the first 7 days free — a valid payment method at signup, cancel anytime. Developer and Studio start at five and ten sites with the developer tooling switched on.
Add your domain in your account portal. The portal shows the exact DNS records for your site; once the domain points here, SSL is issued for it automatically.
Put the site up: upload your application or install your CMS. Managed hosting runs WordPress, WooCommerce, Laravel, other PHP applications and Node applications.
Stuck at any step? Support is on the standard hours in Support & service levels, and the first question is never billed — the included monthly allowance covers getting you unstuck.
Moving an existing site to StackHost without downtime
Set the site up here first and get it fully working before touching DNS. On Developer and above you have a staging environment — rehearse the move there, then repeat it for real.
Switch the domain over last. Change its records to the ones your account portal shows; most updates take effect quickly, though records with a long TTL set by your previous host can linger for a few hours.
Only cancel the old hosting once the site has been served from StackHost for a day or two. Cancelling first is the one mistake that costs you a quiet weekend.
Hands-on migration help — moving files and databases for you — is metered intervention work. It is quoted and approved before it starts, never billed as a surprise. See the intervention terms.
Moving a WordPress or WooCommerce site: what travels with it
A WordPress site is two things: the files (themes, plugins, uploads) and the database (posts, orders, customers, settings). Both have to come across, and the database is the part people forget until the homepage loads and the shop is empty.
The order that works: copy the files and export the database from the old host; bring them across; point the site’s configuration at the imported database; click through every important page on a temporary address; only then switch DNS, as in the moving guide.
WooCommerce adds one rule: take the backup before you touch anything, and do the risky steps in a staging copy first if you are on Developer or above. Stores live and die by their order history.
If you would rather hand us the keys, the move itself is metered intervention work — quoted and approved before it starts. See the intervention terms.
Upgrading or downgrading a plan
Starter to Developer to Studio is a plan change, not a migration. Your sites, domains, SSL and settings stay exactly where they are.
What changes is capacity and tooling: site slots (2 / 5 / 10), the shared envelope (1 / 2 / 4 shared vCPU, 1.5 / 3 / 6 GB RAM, 12 / 30 / 60 GB SSD), and the developer tools — Git, SSH, staging, databases and cron begin at Developer.
Change the plan in your account portal, or write to support if you want a second opinion on which tier fits. See the full comparison on Managed Hosting Managed Hosting.
Names, records and certificates.
Pointing your domain at StackHost
Your account portal lists the exact DNS records for your site. Copy them verbatim into your domain’s DNS settings at your registrar or DNS host.
We do not publish static addresses on this page on purpose: records are tied to your site and can change, and the portal is always current. A guide that quoted addresses here would go stale; your portal cannot.
After changing records, give propagation a little time — how long depends mostly on the TTL your previous host set. Keep the old hosting running until the site answers from StackHost, then cancel it.
SSL certificates: included, renewed, monitored
Every plan includes SSL for your custom domain. The certificate is issued once the domain points at StackHost, and expiry is monitored continuously — renewal is handled for you, not calendar-managed by you.
Seeing a certificate warning? First check the domain’s records still match the portal. If the domain points here correctly and the certificate still misbehaves, that is our side of the line — and problems caused on our side are never billed to you.
The responsibility split between your application and our environment is described on the responsibility boundary.
Running email on your domain alongside your site
If your domain sends and receives mail through an email provider, that provider gives you the records that make it work: MX records for delivery, and SPF, DKIM and DMARC records that tell receiving servers your mail is genuinely yours. Those records live in the same DNS zone as your site’s records.
The two rules that prevent most disasters: add the mail records alongside the ones your StackHost portal shows rather than replacing them, and if you change email providers, update the old provider’s mail records instead of leaving both sets in place — conflicting SPF records are a classic way mail starts landing in spam.
Not sure which records do what? Support can read your current zone and tell you which lines are site and which are mail before you change anything.
DNS changes vs transferring your domain
These are two different things that people often bundle into one worry. Changing DNS records decides where your domain points — it is how your site starts answering from StackHost. Transferring a domain moves who bills you for renewals — the registrar, not the hosting.
You do not have to transfer anything to host with us: point the domain with records and keep the renewal where it is. If you do prefer everything under one roof, a transfer is done after the site is already live here, and it never touches the records themselves.
Whichever you choose, keep the renewal paid. An expired domain takes the site, the mail and the certificates with it, wherever they are hosted.
Propagation, TTL and “why do I still see the old site?”
DNS records carry a TTL — a time-to-live that tells computers everywhere how long to remember the old answer before asking again. When you change a record, machines that recently looked it up keep serving the cached old value until the TTL runs out. That is propagation: not a queue your change sits in, just old answers expiring at their own pace.
Practical consequences: short TTLs make changes take effect quickly but mean more lookups; long TTLs are efficient but slow a switch. Your previous host’s TTL, not ours, decides how long a cutover lingers. And a browser can cache too — testing in a private window or another device rules out your own machine before you blame the internet.
Running sites on the plans.
How plans count sites
Plans count live sites: Starter carries 2, Developer up to 5, Studio up to 10. Each site gets a custom domain and SSL; backups, restore and monitoring are on every plan.
Near the cap and unsure how something counts — an alias, a parked copy, a staging build? Ask before you rely on it. Support will give you a straight answer for your setup rather than a rule that only fits some cases.
The full plan comparison, including what switches on at each tier, is on Managed Hosting Managed Hosting.
Backups and restores
Every plan takes automated backups on its schedule, and monitoring watches the backup jobs where backup is purchased. Studio keeps longer retention than the lower tiers.
Backup storage is also available as an add-on, published in tiers up to 512 GB; larger sizes are quoted per environment.
Restores are a flat-priced operation — the restore fee covers the restore itself and is never multiplied by engineer hours. Request one through support and say which site and which point in time.
Backup tiers and restore terms are published on the pricing page.
Git, SSH, staging and cron (Developer and above)
Developer is where the builder tools begin: Git for deployments, SSH for direct access, a staging environment for testing changes before they ship, databases, scheduled cron jobs and readable logs.
The intended rhythm: build and verify on staging, deploy with Git, read the logs when something looks wrong. Starter does not include these tools — if you are doing real development work, that is the line between the tiers.
Studio adds team access on top, so other people can work without sharing your login.
Cron: work that happens while you sleep
Cron runs commands on a schedule. It is the machinery behind nightly database exports, cleaning out expired sessions, pulling in a feed every hour, or sending the report email at 6am. Developer and above include it, alongside the databases and queues the scheduled work usually talks to.
Two habits keep cron from surprising you: make each job write a log line or leave a file so you can see whether it ran, and never schedule heavy work at the same minute every job on the site fires. A task that used to take seconds and now takes minutes is a quiet signal to look at before it becomes an outage.
What cron is good for and when it starts needing queues rather than schedules — that line is described on the Developer plan page.
When one site outgrows the rest
Managed plans share a resource pool by design — that is what keeps them simple and affordable. If one application grows heavy while the rest stay quiet, upgrading the whole plan for one site’s sake is the wrong move.
The intended exit is a private Stack Server pool: the heavy workload gets reserved capacity of its own, and the quieter sites stay on the plan untouched. Nothing else has to change.
Starting pool sizes are on Stack Servers, and the same logic applies to a single busy application on Cloud VPS.
Keeping a WordPress site healthy
The environment your site runs on is watched, patched and backed up on this side — that is the managed part. The application itself is yours: updates, plugins, passwords and code quality sit on your side of the line, and most WordPress problems start there, not underneath it.
The routine that prevents most incidents: apply WordPress and plugin updates soon after release — in staging first on Developer and above; keep the plugin list short, because every plugin is code you did not write that can update itself; use real passwords and a single admin account per person.
When something breaks, the plan backups mean the question is “what changed yesterday?” rather than “is everything gone?”. Restores are a flat-priced operation, and trouble caused on our side never bills to you.
Backups and restores: the guide above. Where the responsibility line sits: the responsibility boundary.
My site is slow — where do I look first?
Work down this list and you will find the layer that owns the problem:
1. The application. A slow plugin, an unindexed database query or a page calling third-party services explains most slowness. If a plain page on the same site is fast, the environment is not the suspect.
2. The resources. Sites on managed plans share a pool; a burst of traffic or a heavy job can feel like slowness while the envelope refills. If it happens every day at the same time, that pattern is the diagnosis.
3. The lane. A consistently busy site belongs on more capacity — Developer, then Studio, then a private pool or a VPS for the one heavy thing, as in when a site outgrows the plan.
A performance investigation that needs an engineer’s time is metered like any intervention, and approved before it starts.
Choosing and running servers.
Sizing a Cloud VPS
A VPS is priced from what you choose: CPU, memory, storage, operating system and IP addresses. The calculator on the pricing page follows the sliders in real time — nothing is ordered until you say so.
Windows bills vCPU at a different rate than Linux, and a public IPv4 address is part of the default configuration. Start with the shape your workload needs now rather than the one it might need in a year; adjusting later is a conversation, not a rebuild.
The published starting shapes and rates are on Cloud VPS and the rate card in pricing.
Linux or Windows: choosing the operating system
The short version: WordPress, WooCommerce, Laravel, other PHP applications and Node services are at home on Linux, which is also the default. .NET applications and Microsoft database servers are the usual reasons a machine needs Windows.
Both options are fully managed the same way — monitoring, patching and routine maintenance are the operating-system’s job on either. The difference you will see is on the rate card: Windows vCPU bills at its own per-core rate rather than the Linux one, so price the machine in the calculator with the right OS selected before you compare.
With a VPS you hold root on Linux or administrator on Windows either way — the managed layer runs around your access, not instead of it.
What “managed” includes on servers
Every Cloud VPS and Stack Server includes routine infrastructure management: monitoring with alerts, the operating-system baseline and patch cycles, standard security updates and automated maintenance. It is not a line item — it is the product.
Hands-on work is metered and always approved first: manual troubleshooting, recovery, custom firewall or runtime work, migrations. It is billed at $35.00 per hour in 30-minute blocks, drawn from 30 included minutes per account per month, then any prepaid pack.
The full boundary — included, metered, and the pre-approval rule — is on Support & service levels.
Stack Server pools: one pool, several machines
A Stack Server pool is reserved CPU, memory and storage that you shape into one or more virtual machines. S, M and L are starting sizes (2 vCPU · 4 GB · 80 GB, 4 · 8 · 160, and 8 · 16 · 320); anything else is calculated from the same published resource rates.
Several environments can share one pool — a common shape is production on one machine and staging on a smaller one, drawing from the same reserved envelope. Machines inside a pool share its capacity, so size the pool for the sum of what runs on it.
Pools are managed compute with reserved resources, not single-tenant hardware — the distinction is spelled out on Stack Servers.
Snapshots and backups are different things
A snapshot is a point-in-time copy of a machine’s storage, taken before risky work — an upgrade, a configuration change — so you can step back if it goes wrong. It is short-term insurance with a mechanical trigger: you take it when you are about to do something.
A backup is the scheduled, ongoing copy of your data that survives the loss of the machine itself. Backups run on the plan schedule with retention, and restores are a priced operation. On Cloud VPS a snapshot bills at $2.57 plus $0.0391 per GB — one full copy of the provisioned storage; Stack Server pool snapshots are quoted with your pool.
They answer different accidents. The snapshot undoes Tuesday’s upgrade; the backup recovers from losing the machine. For work you cannot afford to lose, you want both.
Snapshot options appear in the calculators; backup tiers are on the pricing page.
Money, hours and how to reach us.
Understanding your bill
Managed Hosting bills fixed Kenyan shilling plan prices — KSh 850, KSh 1,699 or KSh 4,500 a month. Cloud VPS bills the published monthly price of the size you ordered, and Stack Server pools bill their starting price plus any resources added at the published US dollar rates; any KSh or ZAR figure shown beside them on the site is an indicative conversion, and the invoice confirms the billed amount and currency.
Support packs are priced in US dollars and convert at the indicative rate shown when you buy. Backup storage and restore operations bill at their published prices. There are no setup fees hidden in the fine print — the rates on the pricing page are the rates.
Getting support: hours, after-hours and packs
Monitoring and automated responses run around the clock on every plan. Engineer support keeps published hours — Monday to Friday, 08:00–18:00 EAT, Kenyan public holidays excepted — and work outside those hours still happens; it consumes support time at twice the rate.
Every account carries 30 included intervention minutes a month. Prepaid packs (5, 10 or 20 hours) belong to the account rather than one server and last 12 months. Before any billable work starts, you approve it — including a spending cap for emergencies if you want one.
The whole model, with the no-surprises rules, is on Support & service levels.
What gets confirmed before you order
The calculators show what the published rules produce. Anything outside those rules — a custom server size, additional IP addresses, longer backup retention, a special support arrangement — is confirmed with you before the order, never assumed into it.
The same logic runs the other way: if work would bill, you approve it first, and if a StackHost change caused the problem, correcting it never touches your allowance or a pack balance.
Write to [email protected] with what you need; a human answers with the terms in writing.
Support packs: how they are used up
A pack is prepaid intervention hours that belong to the account, not to one server — every machine and pool you run draws from the same balance. Consumption follows a fixed order: the 30 included minutes each month go first, then the pack, then pay-as-you-go at the standard hourly rate.
After-hours work consumes time at twice the rate — one hour outside support hours uses two — from whichever balance covers it. A pack lasts 12 months from purchase, and one extension is granted if you top up before it expires. Unused hours are not cash-refundable.
Every draw is visible: billable work is described and approved before it starts, so the balance should never move without you knowing why.
Pack prices and the full terms: the support section on pricing and Support & service levels.
Where to ask what
Before you order — custom sizes, extra addresses, retention, availability commitments in writing: [email protected]. You get the terms in writing, and they are the terms the order carries.
How do I…? — start here in the knowledge base; if the guide leaves a gap, say which one and it gets filled.
Something is wrong right now — write to support. Monitoring is already watching the platform around the clock and a failed check pages an engineer before your email arrives; your message adds the part only you can see, like which page and when.
Your data — copies, corrections and deletions follow the privacy page: one email, handled by a person.