Server operations
Request IPMI or KVM console access
Request console access for boot and connection problems, then verify the server before making changes.
On this page
Before you start
- A signed-in account containing the target server.
- The server name, region, and the approximate time the problem began.
- A clear reason for console access, such as a boot failure or lost network connectivity.
- A secure workstation and a plan to close the console as soon as diagnosis is complete.
Use the console for out-of-band diagnosis
IPMI/KVM is useful when the operating system is not reachable over SSH: for example, while observing firmware startup, selecting a boot device, reading an early kernel error, or repairing a network configuration. It is not the normal administration channel and should not be left open for routine shell work.
The KVM request is associated with a server owned by the signed-in account. The service accepts the request, creates an action reference, and reports QUEUED. That response confirms storage of the request only; it does not prove that a management controller was contacted or that a session exists.
Queue the request
Do not expect a browser window to open. There is no GET action route in the API, so refreshing the dashboard cannot show progress or completion. Preserve the visible action suffix in case the operator needs to correlate the request with the queue.
- Open Account > Servers and locate the correct server by name and region.
- Confirm that the state and timing match the incident you are investigating.
- Select Open KVM once. The button changes to show a shortened queued action reference.
- Copy the full context you have: server name, region, order reference if available, time, and the shortened action reference.
- Contact the operator through the documented support channel and request fulfilment of that queued action.
Handle delivered console access safely
- Verify that the operator’s message identifies the expected server and that the console endpoint uses HTTPS.
- Treat a console link or credential as a secret. Do not paste it into chat, logs, screenshots, or shared issue trackers.
- Observe first. Avoid changing boot order, firmware, storage controllers, or RAID settings unless the recovery plan requires it.
- Never initialize, format, or rebuild a disk before verifying backups and the exact target device.
- Close the browser session and ask for revocation when the work is complete.
Expected result and escalation
The immediate expected result in the account is a queued action reference. The operational result, available only after operator fulfilment, is a temporary remote-console session connected to the intended server. Use it to determine whether the failure occurs before boot, during boot, or after the operating system starts.
If the request is queued but no access arrives, do not generate many duplicate KVM requests. Send the operator one concise report containing the server, region, action suffix, request time, and business impact. If a console opens on an unexpected machine, disconnect without interacting and report it as a security issue.
| Observation | Likely boundary | Next step |
|---|---|---|
| No console URL after queueing | Control-plane limitation | Ask the operator to fulfil the queued request. |
| Firmware visible, OS does not boot | Boot or storage issue | Capture the exact error and follow recovery guidance. |
| OS login appears, SSH still fails | Guest network, firewall, or SSH service | Inspect configuration from console before changing it. |
| Wrong server displayed | Access-assignment error | Disconnect and report immediately. |