Hello Don,
I had a missunderstanding on my side, I assumed that this processtype is related to the "users point of view" and indicates and Users "play" as edge and internal (i.e. publish) as origin. Thanks a lot for sorting this out.
I try to "re validate" my findings with your statement and have another trouble (or again a missunderstanding).
When I start to play multiple times the same stream (using each time a different streamaliases) I see multiple INR/edge created, each look related to an ONR/origin. In total I have same amount of ONR/origin, INR/edge as number of corse. Which looks fitting your explanation.
Each INR/edge have USers "play" ONR/edge associated which looks well balanced.
In a tree it would looks like
StreamID=1 INR/origing (RTMP ingest/publish)
StreamID=12 ONR/origin (feed for edge process)
StreamID=67289252626438 INR/edge (feed from origin process)
StreamID=67289252626437 ONR/edge # USers "play"
StreamID=67289252626439 ONR/edge # USers "play"
The Usersplay Streams (ID 67289252626437, 67289252626439) have a StreamALias information and Users IP, which sounds fully correct.
The ONR/origin (ID12) have as well a Streamalias information (the Streamalias from first user request), this sounds not understandable for me. The INR/edge (ID 67289252626438) have the streamalias "included in the pull request uri", which looks confusing as the alias is created for the user and not for the edge/origin.
If i like to shutdown a Stream (defined by streamalias) the first match (when parsing all streams) will be the ONR/origin, as it´s id is lowest. If i shutdown this stream i will shutdown all related childs which means in the example above as well another stream (67289252626439) with a different Streamalias.
In my understanding a streamalias is related to a usersrequest and should not be visible on internal connections/streams required to keep origin/edge feeded. listing/showing this data on the streams between origin-edge it sounds a risk to shutdown multiple streams instead of one. Anyway if checking additional the "users IP" it will be possible to develop a workaround…
Kind Regards,
Uwe