Add logging for high keyframe interval, reduce log level for discovery loop

[eliteprox]

What does this pull request do? Explain your changes. (required)

Specific updates (required)

• Added logging to SegmenterTimeout error in media_server.go to suggest for the broadcaster to lower their keyframe interval. Confirmed this error often occurs when the keyframe interval is 8 seconds or higher.

  • RTMP Timeout: Ensure keyframe interval is less than 8 seconds

• Reduced log level for orchestrator discovery loop during streaming to debug, this helps prevent excessive logging when broadcaster has only a few orchestrators. Example of issue: Broadcaster Discovery Loop Errors.txt

How did you test each of these updates (required)

1. Broadcasted with and without -orchAddr using low deposit.
2. Changed log level with -v flag. Confirmed discovery loop error logs will appear only if -v 5 or higher is used.
3. Captured log behavior while streaming New Broadcasting Logs.txt

Does this pull request close any open issues?

None

Checklist:

• Read the contribution guide
• make runs successfully
• All tests in ./test.sh pass
• README and other documentation updated
• Pending changelog updated

[codecov]

Codecov Report

Merging #2709 (9c91d1d) into master (cf95f00) will decrease coverage by 0.01929%.
The diff coverage is 10.00000%.

@@                 Coverage Diff                 @@
##              master       #2709         +/-   ##
===================================================
- Coverage   56.35194%   56.33265%   -0.01929%     
===================================================
  Files             88          88                 
  Lines          19128       19131          +3     
===================================================
- Hits           10779       10777          -2     
- Misses          7761        7765          +4     
- Partials         588         589          +1     

Impacted Files	Coverage Δ
cmd/livepeer_cli/wizard_stats.go	0.00000% <0.00000%> (ø)
server/mediaserver.go	67.35632% <0.00000%> (-0.20503%)	⬇️
discovery/discovery.go	91.66667% <100.00000%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 82e40f9...9c91d1d. Read the comment docs.

Impacted Files	Coverage Δ
cmd/livepeer_cli/wizard_stats.go	0.00000% <0.00000%> (ø)
server/mediaserver.go	67.35632% <0.00000%> (-0.20503%)	⬇️
discovery/discovery.go	91.66667% <100.00000%> (ø)

[Titan-Node]

Reduced log level for orchestrator discovery loop during streaming to debug, this helps prevent excessive logging when broadcaster has only a few orchestrators. Example of issue: Broadcaster Discovery Loop Errors.txt

By default a B will timeout all segments received equal or greater than 8 seconds.
We tried to find the variable for the RTMP timeout to allow OBS to send segments greater than 7 seconds (I think OBS defaults to 8 seconds) but couldn't find it. Unless you manually change OBS keyframe interval to less than 7 seconds it does not work. Hopefully this log will save people a lot of time.

Added logging to SegmenterTimeout error in media_server.go to suggest for the broadcaster to lower their keyframe interval. Confirmed this error often occurs when the keyframe interval is 8 seconds or higher.

Works great, really needed to clean up the logs while using the -pricePerBroadcaster set to 0 for free transcoding.

[AlexKordic]

Looks good

[cyberj0g]

LGTM!