All paths on this page start with /services/virtual-servers. A server is named by its id. A server that is not yours, or that is deleted, is 404 not_found.
The server object
Every endpoint that shows or changes a server answers with this object.
{
"id": 219,
"name": "web-1",
"status": "active",
"plan": {"id": 3, "name": "Core-4"},
"location": {"id": 4, "code": "EYG-DC1", "name": "Netherlands DC1", "country_code": "NL"},
"os": {"code": "ubuntu-server-24-04", "name": "Ubuntu Server 24.04 LTS", "family": "linux"},
"resources": {"cpu_cores": 4, "ram_mb": 8192, "storage_gb": 120},
"ipv4": [
{"address": "203.0.113.10", "gateway": "203.0.113.1", "primary": true, "addon_id": null},
{"address": "203.0.113.25", "gateway": "203.0.113.1", "primary": false, "addon_id": 341}
],
"ipv6": [
{"prefix": "2001:db8:10:2a::/64", "gateway": "2001:db8:10::1", "enabled": true}
],
"billing": {
"period": "monthly",
"price": "21.97000000",
"currency": "EUR",
"auto_renewal": true,
"next_billing_at": "2026-12-12T16:52:06Z",
"expires_at": "2026-12-12T16:52:06Z"
},
"addons": [
{"id": 341, "name": "Additional IPv4", "type": "additional_ipv4", "status": "active", "price": "2.00000000", "billing_period": "monthly", "next_billing_at": "2026-12-12T16:52:06Z"}
],
"notes": "Production web server",
"created_at": "2026-11-12T16:52:06Z"
}| Field | Meaning |
|---|---|
name |
a label of your choice, not the server's hostname |
ipv4[].primary |
the main address, configured in the server automatically |
ipv4[].addon_id |
the paid add-on that holds an additional address; null for the main address and for addresses from your own BYOIP pool |
ipv6[].enabled |
false when IPv6 is switched off: the prefix stays reserved for the server |
billing.price |
what one billing period of the server costs, without add-ons |
billing.next_billing_at |
when the next period is due: the server is renewed then, or it expires |
billing.expires_at |
when a server that was not renewed is deleted for good, see When a server is not renewed |
addons |
paid extras renewed together with the server; type is additional_ipv4 or backup_daily |
notes |
your own notes, up to 1000 characters |
Statuses
| Status | Meaning |
|---|---|
ordered |
just ordered, waiting to be built |
pending |
being built; it becomes active when it is ready |
active |
running and paid for |
reinstalling |
the OS is being reinstalled; it becomes active again |
expired |
not renewed in time: the server is stopped. Renew it before billing.expires_at, otherwise it is deleted for good |
suspended |
suspended by us, for example over an abuse report; contact support |
failed |
creating or reinstalling the server failed, and it cannot be used. Contact support with its id |
terminated |
deleted. Shown only once, in the answer to POST /{id}/terminate; after that, and after an automatic deletion, the id answers 404 not_found |
Most actions need the server to be active; otherwise the answer is 409 invalid_state.
Reading
Ability: vps:read.
| Request | What it returns |
|---|---|
GET / |
all your servers except deleted ones, as a list |
GET /{id} |
one server |
GET /{id}/status |
the live state of the virtual machine |
GET /{id}/stats |
load graphs |
GET /{id}/traffic |
collected traffic |
GET /{id}/activity |
the server's activity log |
GET /{id}/guest-agent |
whether changes can be applied without a reboot |
GET /{id}/rdns |
reverse DNS names of the IPv4 addresses |
GET /{id}/billing-periods |
periods the server can be paid for, with prices |
GET /{id}/os-templates |
systems the server can be reinstalled with |
GET /{id}/backups |
backups and the backup operation in progress |
GET /{id}/ssh-keys |
the SSH keys the panel manages on the server |
Live status
GET /{id}/status{"data": {"status": "running", "lock": null, "uptime": 86400, "cpu": 0.0315, "memory": {"used": 1073741824, "max": 8589934592}}}status is running or stopped, or unknown when the state could not be read right now: ask again a little later. lock names an operation that holds the machine right now (for example backup), during which most actions are refused. uptime is in seconds, cpu is the load as a fraction of the server's cores (1 = all cores busy), memory is in bytes.
Graphs
GET /{id}/stats?timeframe=hourtimeframe is hour (default), day, week, month or year. The answer is {"data": {"timeframe": "hour", "points": [...]}}; each point has time (Unix timestamp), cpu (fraction), mem and maxmem (bytes), netin and netout, diskread and diskwrite (bytes per second).
Traffic
GET /{id}/traffic?from=1762905600&to=1765497600&granularity=day| Parameter | |
|---|---|
from, to |
required, Unix timestamps; to is cut at the current hour |
granularity |
day (default), hour (a range of 31 days at most) or summary |
{"data": {"granularity": "day", "from": 1762905600, "to": 1765497600, "total": null, "points": [{"bucket": 1762905600, "bytes_in": 52428800, "bytes_out": 157286400, "bytes_total": 209715200}]}}With summary, points is empty and total holds one object with bytes_in, bytes_out and bytes_total.
Activity
GET /{id}/activity?page=1Newest first, 10 entries a page. Each entry has id, action, description, category, status, error_message, ip_address, user_name and created_at. meta has current_page, last_page, per_page and total.
Power
POST /{id}/powerAbility: vps:power.
{"action": "reboot"}| Action | What it does |
|---|---|
start |
power on |
shutdown |
ask the system to shut down cleanly |
stop |
power off at once, like pulling the plug |
reboot |
ask the system to restart cleanly |
reset |
hard restart at once |
The answer {"data": {"action": "reboot"}} means the command was accepted. Follow the result with GET /{id}/status.
Settings
Ability: vps:manage. Each answers with the server object (rDNS: with the rDNS list).
| Request | Body |
|---|---|
PATCH /{id}/name |
{"name": "web-1"}: 3 to 50 characters; letters, digits, spaces, dots and hyphens, starting and ending with a letter or digit |
PATCH /{id}/notes |
{"notes": "..."}: up to 1000 characters, null clears them |
PATCH /{id}/rdns |
{"ip": "203.0.113.10", "ptr": "mail.example.com"}: ip is one of the server's IPv4 addresses, ptr: null removes the record |
The rDNS list, of GET /{id}/rdns and of the change, has one entry per IPv4 address: {"data": [{"ip": "203.0.113.10", "ptr": "mail.example.com"}]}, with ptr: null where no name is set.
Password and SSH keys
Ability: vps:manage.
Each change can be applied two ways:
- The plain path stores the change at once; the server picks it up at the next reboot. The answer has
"requires_reboot": true. - The
.../agentpath applies it inside the running server at once, through the QEMU guest agent. The answer has"requires_reboot": false. CheckGET /{id}/guest-agentfirst:{"data": {"available": true}}.
| Request | Body |
|---|---|
POST /{id}/password |
{"password": "..."} |
POST /{id}/password/agent |
{"password": "..."} |
POST /{id}/ssh-keys |
{"ssh_keys": ["ssh-ed25519 AAAA... me@laptop"]} |
POST /{id}/ssh-keys/agent |
{"ssh_keys": [...]} |
- A password has at least 10 characters with a lower-case letter, an upper-case letter and a digit. A password is never sent back by the API.
- SSH keys are always sent as the whole set: adding and removing a key are the same operation. Read the current set with
GET /{id}/ssh-keys, change it, send it back. The answer includes the new set. - Up to 10 keys, one key per list item. Accepted types:
ssh-ed25519,ecdsa-sha2-nistp256/384/521andssh-rsaof 2048 bits or more. - An empty list removes all keys, and is accepted only while the server has a password: otherwise you would lock yourself out.
- The
agentpath leaves alone keys you added inside the server by hand. - SSH keys are not supported on Windows servers. Keys can be saved 10 times an hour per server, both ways together.
{"data": {"requires_reboot": true, "ssh_keys": [{"type": "ssh-ed25519", "comment": "me@laptop", "fingerprint": "SHA256:mVPwvezndPv/ARoIadVY98vAC0g+P/5633yTC4d/wXE", "key": "ssh-ed25519 AAAA... me@laptop"}]}}Network
An address is always named by the IP itself, as in ipv4[].address of the server object. Each change answers with the server object and a meta block about what happened. Network changes of one server run one at a time: a second one started during the first gets 409 busy.
| Request | Ability | Body | What happens |
|---|---|---|---|
POST /{id}/change-ip |
vps:billing |
{"ip": "203.0.113.10"} |
the address is replaced with a new one. A fixed fee is charged; free when the new address comes from your own BYOIP pool |
POST /{id}/primary-ip |
vps:manage |
{"ip": "203.0.113.25"} |
another address of the server becomes the main one. Free |
PATCH /{id}/ipv6 |
vps:manage |
{"enabled": false} |
IPv6 is switched off or back on; the prefix stays with the server. Free, reboots the server |
POST /{id}/additional-ipv4 |
vps:billing |
none | one more IPv4 is added |
POST /{id}/additional-ipv4/{addon_id}/cancel |
vps:billing |
none | a paid additional address is given up |
meta of an IP replacement:
{"meta": {"old_ip": "203.0.113.10", "new_ip": "203.0.113.77", "charged": "1.00000000", "currency": "EUR", "rebooted": true}}- Replacing the main address, changing the main address and switching IPv6 reboot the server to apply the change;
meta.rebootedtells whether this change did. Replacing or adding an additional address does not: such an address is not configured in the server automatically, set it up in the system yourself. - An additional IPv4 is an add-on: the rest of the current billing period is charged at once, then the address renews with the server.
metahas the newip, itsaddon_id(nullwhen it came free from your BYOIP pool),chargedandcurrency. - Cancelling an additional address takes
addon_idfrom the server object. It happens at once and is not refunded.
Console
POST /{id}/consoleAbility: vps:manage.
What a noVNC client needs to show the server's screen:
{"data": {"websocket_url": "wss://panel.hostoff.net/api/vortex/wss/...", "password": "..."}}- The connection is made for this call and must be opened within about 10 seconds. Request it at the moment a user opens the console, and request it again to reconnect.
- In a browser with noVNC 1.4 or later:
new RFB(element, data.websocket_url, { credentials: { password: data.password } }). The WebSocket goes through our domain, so the page itself can be on yours. - The screen of a server is full access to it. Do not store the answer and do not pass it to anyone but the user who opened the console.
- At most 30 consoles an hour per server and 5 open at the same time.
Backups
The platform makes scheduled backups by itself, and the list is always visible. Creating, restoring and deleting backups needs the backup add-on on the server; without it these requests answer 409 not_supported.
GET /{id}/backups{
"data": {
"can_manage": true,
"addon": {"status": "active", "price": "4.39000000", "currency": "EUR", "paid_until": "2026-12-12T16:52:06Z"},
"addon_offer": null,
"active_task": null,
"last_task": {"id": 77, "action": "create", "backup_id": null, "status": "completed", "failure_reason": null, "created_at": "2026-11-14T10:00:00Z", "finished_at": "2026-11-14T10:03:12Z"},
"backups": [
{"id": "2026-11-14T10:00:05Z", "type": "manual", "size": 2147483648, "verification": "ok", "protected": false, "created_at": "2026-11-14T10:00:05Z"}
]
}
}| Field | Meaning |
|---|---|
can_manage |
whether backups can be created, restored and deleted now |
addon |
the server's backup add-on, when it has one |
addon_offer |
what buying the add-on costs: price per billing_period and first_charge, the part charged now for the rest of the current period, in currency. null when it is not offered or already bought |
active_task, last_task |
the operation in progress, and the last one that finished |
backups[].id |
the name of the backup in the requests below |
backups[].type |
auto for scheduled backups, manual for the ones you made |
backups[].verification |
the result of the integrity check (ok, failed), null while not checked |
backups[].protected |
a protected backup cannot be deleted: the answer is 409 invalid_state |
| Request | Ability | What it does |
|---|---|---|
POST /{id}/backups |
vps:manage |
create a backup now |
POST /{id}/backups/{backup_id}/delete |
vps:manage |
delete a backup |
POST /{id}/backups/{backup_id}/restore |
vps:destroy |
restore the server from a backup |
POST /{id}/backups/addon |
vps:billing |
buy the backup add-on; meta has addon_id, charged, currency |
POST /{id}/backups/addon/cancel |
vps:billing |
cancel the add-on at once, without a refund; existing backups stay listed |
- An operation is queued and the answer is its task. While it runs,
GET /{id}/backupsshows it asactive_task; then aslast_task. A task'sstatusisqueued,running,completedorfailed. - Only one backup operation runs at a time per server: a second one gets
409 busy. - Restoring hard-stops the server and overwrites its disks: everything that changed after the backup is lost. That is why it needs
vps:destroy. - Per server and hour: 2 backups by hand, 3 restores, 10 deletions.
Renewal and billing period
Ability: vps:billing. Money is charged from the account balance; amounts are never sent in requests.
| Request | Body | What it does |
|---|---|---|
POST /{id}/renew |
none | pays for one more period |
PATCH /{id}/auto-renewal |
{"auto_renewal": true} |
switches automatic renewal on or off |
POST /{id}/change-billing-period |
{"billing_period_id": 5} |
moves the server to another billing period |
- Renewal pays for the server and its add-ons together, and
meta.chargedis the total. You can renew ahead of time and more than once: each call adds a period. Renewing anexpiredserver makes itactiveagain and starts it. - Changing the period charges nothing now: the new period and its price apply from the next renewal. Take the id from
GET /{id}/billing-periods, which lists each period'sprice,addons_price,total_price,is_currentandis_available. It is not possible for a server with an individual price or configuration (can_change_period: false), nor to a period on which one of its add-ons is not sold: both answer409 not_supported. A period with no free server slot answers409 limit_reached(seebilling_limitsinGET /account); a period the server is already on, or one not sold for its plan and location, answers422 invalid_input.
When a server is not renewed
At billing.next_billing_at a server with auto-renewal on is renewed from the balance, together with its add-ons. If auto-renewal is off, or the balance does not cover the server and its add-ons:
- The server becomes
expiredand is stopped. Its disks and addresses are kept for now. - Until
billing.expires_at,POST /{id}/renewbrings it back: it becomesactiveand is started. Reading and deleting still work; every other action answers409 invalid_state. The grace period depends on the billing period: currently 24 hours for hourly servers, 48 hours for daily and 3 days for longer periods. - At
billing.expires_atthe server is deleted for good, the same asPOST /{id}/terminate: its data and add-ons are gone, its addresses are released, nothing is refunded. It cannot be restored.
To keep a server, keep auto-renewal on and enough money on the balance, or renew it ahead of time.
Reinstall
POST /{id}/reinstallAbility: vps:destroy.
{"os_template_id": 1, "password": "Str0ngPassw0rd", "ssh_keys": ["ssh-ed25519 AAAA... me@laptop"]}- Everything on the server's disk is erased. The addresses and resources stay.
- Take
os_template_idfromGET /{id}/os-templates: only the systems available for this server are listed. - Give a password, SSH keys, or both. The same rules apply as in Password and SSH keys.
- The answer is the server in the status
reinstalling; it becomesactiveagain when the system is ready. At most 3 reinstalls an hour per server.
Delete
POST /{id}/terminateAbility: vps:destroy.
- The server, its data and its add-ons are deleted for good and its addresses are released. Deletion cannot be undone.
- Nothing is refunded, not even the unused part of the paid period.
- Possible for servers in the status
active,expiredorsuspended. A server that isordered,pending,reinstallingorfailedanswers409 invalid_state. - The answer is the server in the status
terminated, the last time it is shown: from then on it is not in the list, and its id answers404 not_found.
Neither reinstalling nor deleting asks for a confirmation in the request: the call with a token that has vps:destroy is the confirmation. Give that ability only to the tokens that need it.