HTTP streams fail after encoder reconnect due to network issues
We observed several times in cases when the network connection from the encoder to the server was disrupted shortly that the RTMP streams will continue however the HTTP streams for cupertino and sanjose did not work any more.
We use FMLE as encoder and it automatically reconnects to the server:
Take a look at the Wowza logs for more information. You may see that the HTTP streams have been disconnected, since the time the stream was dropped is longer that the timeout Wowza has for holding the connections open.
For Flash, there is a robust player built around the the stream. For Cupertino, we don't have this, but it is possible to create such a player that can monitor the stream, respond to disconnect events, and reconnect as you desire. See the Apple resources for examples:
For San Jose you can also check the Wowza OSMF example player code and monitor for events. These are the suggestions I have. There may be others.