Closed victorges closed 3 months ago
@victorges I've done a refactor of this now so that we only check the backup location at the beginning of the job, working out which manifest to use and writing out a new manifest with absolute URLs if necessary. Then probing, clipping etc works as before (in theory). Does it look ok to you so far?
@victorges please could you give this one more sign off?
This implements the storage fallback logic following the design doc.
This builds on top of https://github.com/livepeer/go-tools/pull/121 and https://github.com/livepeer/catalyst-api/pull/1302
The trickiest bits were the error handling for both manifest and segment file reading. We need to handle not found errors explicitly from primary and backup to return appropriate errors to the caller.