Monitoring
Watching a running show, and finding out why something looks wrong. This is the stage you return to during a run rather than one you pass through once.
Do this in order
-
Read the Performance Overview — frame timing, GPU memory, per-stream playback and output status, per render group. The pre-show health check.
-
Check the output signal status — per output, whether the link is up and whether it has the bandwidth for the mode being asked of it. This lives on the server side, in the Hub.
-
Read the Log — what the server reports about itself, with per-severity filtering, search, and imported logs for comparing sessions.
Done when: you can tell, before an audience arrives, whether the show is running cleanly — and when it is not, which layer is at fault.
What a warning means, and what an error means
Backstage's log severities are not decoration. The distinction is deliberate and it tells you whether to act:
|
Severity |
What it means |
|---|---|
|
Warning |
An expected operational state — usually something this machine does not have. Media a project references but this server has not mounted, an audio device the project names but the machine lacks, ASIO channels a device does not expose, a UDP peer that is not up. Nothing is broken. A project built for another machine will produce these by design. |
|
Error |
A program fault — something that should have worked and did not. These are worth chasing. |
Two consequences when reading a log:
-
A wall of warnings on project load is normal when the project came from a different server. Look at what is missing, not at the count.
-
One fault is reported once. Backstage bails at the first failure rather than continuing with a null result and reporting again, so a single cause does not appear as a chain of unrelated errors. If you see many distinct errors, they are many distinct problems.
The default log level is Notice, which hides Info, Debug and Trace. If a message you expect is not there, lower the level before concluding it was never logged.
When something is wrong
Monitoring tells you that something is off. Troubleshooting covers the specific symptoms — black output, assets not loading, connections not firing, servers out of sync.
is currently filed under Monitoring. Replicating assets between servers is not monitoring — it belongs in Content. Move it, or say where it should go.]