I have the task of restreaming a RTMP stream that exists on a CDN. We currently pull the stream with streammanager and then make it available as HLS. However, we have multi-server environment and would like to build in some redundancy on the mechanism which pulls and segments.
If we were to use two origins ( which is the normal approach) then each origin produces
differently named .ts files
splits the segments at different keyframes.
W are using the HTTP cache origin feature to remove the ?wowzasessionid so that the resultant segments can be CDN cached in return.
Is there a way of synching and asking for similar names across the origins or another way to build redundancy?
There is nothing built-in to Wowza to sync servers in this way. You can configure primary and backup origins using pipe delimited urls in your .stream file, but the failover happens when a client connects, and this is not going to work for HTTPOrigin server because clients playback the CDN cache not from Wowza.
You can sync chunks from the same source in two servers using the Application.xml /LiveStreamPacketizer cupertinoCalculateChunkIDBasedOnTimecode Property
You can sync chunks from the same source in two servers using the Application.xml /LiveStreamPacketizer cupertinoCalculateChunkIDBasedOnTimecode Property
I’ve been trying to replicate this all day with no joy - I’ve added it into the /conf/livecf/Application.xml file from the new(ish) CloudFront distribution packages and it’s not creating the chunk Id based on timestamp - which is a problem