Debugging partial pipeline stall?

Hello everyone,

I have a rather complex pipeline, with several WebRTC endpoints that can 
dynamically connect and disconnect, thereby also influencing other internal bits 
of the pipeline plumbing.

Generally, this works fine, but I do still have occasional hangs where some 
parts of the pipeline just stop delivering data downstream. This occurs 
_sometimes_ after disconnecting a client - I've already used the leaks tracer to 
verify that I'm not leaving any dangling pads or similar floating around. The 
pipeline graph in that semi-stalled state also looks perfectly normal.

I'm a bit stumped about how to proceed further. Looking at GST_DEBUG at various 
levels hasn't gotten me anywhere either (maybe due to sheer volume). Any 
suggestions for debug filters or tracers I could use here, and also ideas/tools 
for parsing the massive amount of output?

Thanks and best, Florian
-- 
SENT FROM MY DEC VT50 TERMINAL