Server sleep, wake and availability
System sleep suspends ordinary server execution, and recovery after wake must restore the process's network and storage environment before service is fully available.
When a server enters a system sleep state, ordinary application execution is suspended according to that state's rules. Emby cannot continue its normal playback, API handling or background tasks as if the system were fully active.
This is different from one disk spinning down, where the operating system can remain awake and only that device needs to resume.
Which part stops?
| Condition | What changes |
|---|---|
| Individual drive standby | First access to that drive can need a wake interval |
| System suspend | Ordinary server work stops until resume |
| Hibernation | System state is stored for later restoration under the platform's rules |
| Shutdown | A new startup is needed |
| Platform-specific low-power idle | Capabilities depend on the platform's documented state |
The names and implementations differ between operating systems. A modern low-power state should not be assumed to behave identically to every traditional suspend state.
Wake capability has a path
Wake-on-network behavior depends on hardware, firmware, operating-system policy and which sleep state is in use. The network also has to deliver the applicable wake signal to the machine.
A server that can wake from a local test does not thereby have an internet-reachable wake mechanism. Discovery, routing and any forwarding arrangement have their own scopes.
After wake, an existing process may resume, but its earlier connections can be invalid. Address changes, disconnected shares and device reinitialization can affect the operation it performs next.
Availability returns in stages
An original resume sequence is:
System resumes
Network interface becomes usable
Expected media share becomes readable
Server operation succeeds under its service identity
Those observations need not happen at the same instant. A configured library path can exist before the intended remote tree is actually accessible.
A task scheduled during suspension also did not execute its normal work at that time. Any catch-up behavior belongs to the scheduler and task policy.
For an Emby host, establish the availability expected during quiet periods and after resume. A visible process or responding host is only part of the evidence that the required media operation can now complete.
See Also
- Disk spin-down, wake latency and idle workloads
- Background task scheduling and shared resources
- Emby server discovery and advertised endpoints
- Unavailable network mounts and trustworthy library state