Health signals explain whether the host can sustain the media workflow. Use them with module statistics: a healthy host does not prove a source is live, and a live source does not prove the host has headroom.
開始之前
A known normal baseline for this instance and workload.
Permission to inspect or clear operational error logs.
An external incident record before destructive cleanup.
設定說明
只說明操作員需要的控制項；內部欄位名稱與實作事件不會顯示。
Capacity
Read current host headroom.
- CPU usage
Current processing load across media and application work.
- Memory usage
Current RAM pressure.
- Disk usage
Space available for logs, uploads, temporary transcoding and recordings.
- Network speed test
Measured network capacity from the instance.
Run outside critical production peaks.
Operational evidence
Identify failures before changing configuration.
- Process errors
Application/media process failures requiring module-level investigation.
- SRT errors
SRT transport events and errors associated with contribution workflows.
- Version notification
Installed application version and update context.
- Clear error logs
Remove accumulated error entries.
Export or record incident evidence first.- IP location
Resolve regional context for a connection address when supported.
安全的首次工作流程
- 01
Compare CPU, memory and disk with the known healthy baseline.
- 02
Open process or SRT errors and identify the owning module and time window.
- 03
Inspect that module's live statistics and real input/output.
- 04
Run a speed test only when it will not distort production traffic.
- 05
Record or export incident evidence before clearing logs.
工作流程範例
Diagnose unstable live transcoding
Output frames drop after a new transcoding job starts.
如何建立
- 1
Confirm the source bitrate is stable.
- 2
Compare host CPU/memory before and after the job starts.
- 3
Inspect output bitrate and frame rate.
- 4
Remove unnecessary resize/frame-rate/codec work or move to measured host capacity.
Preserve evidence before clearing errors
Old errors obscure an active incident investigation.
如何建立
- 1
Record the relevant error, timestamp, module and current release version.
- 2
Save the associated module configuration and live statistics.
- 3
Clear errors only after the evidence is externalized.
- 4
Reproduce one controlled test and investigate only the new entries.
驗證結果
- Capacity values match the expected workload and leave planned headroom.
- Every active error can be tied to a time and owning module.
- Clearing logs occurs only after evidence is stored elsewhere.
疑難排解
Host health is normal but playback fails
檢查- Inspect the source and output modules.
- Check access, protocol and real client playback.
Treat host health as one layer; repair the first failed media boundary.