Alternate BaseURL failover broken

[davemevans]

It looks like 636f4f9 has broken BaseURL failover.

This appears to be because the scheduler is not running when a failure occurs, and this is required in order to re-request segments from an alternative BaseURL. It used to be the case that replacement requests were downloaded immediately but this seems to have been replaced with a queue which is serviced by a scheduler.

More as we have it.

#1701 is also required.

[davemevans]

I think we have a fix which could be PR'ed tomorrow after some internal review.

[dsparacio]

the Replacement queue was put in for fast switching in May. Can we not invoke the scheduler method outside of the timer based on some failover signal or error?

[davemevans]

Can we not invoke the scheduler method outside of the timer based on some failover signal or error?

Pretty much, yes. Needed a further tweak to ensure initialisation segments were handled correctly.