Watch depends on long running connections, which intervening proxies
may break without the control of the remote server. Specific errors
handled are io.EOF, io.EOF wrapped by *url.Error, http connection
reset errors (caused by race conditions in golang http code), and
connection reset by peer (simply tolerated).
Watches are often established via long-running HTTP GET requests which
will inevitably time out during the normal course of operations. When
the watches time out, an io.EOF or an io.ErrUnexpectedEOF will bubble
up to client components such as StreamWatcher and Reflector. Treat EOF
as a clean watch termination. Treat ErrUnexpectedEOF as a less-clean
but non-fatal watch termination and log the event at the warning level.
This greatly reduces the amount of log noise generated during what is
ultimately normal operation, and adds the flexibility for the operator
to make a distinction between the EOF conditions if so desired (by
adjusting the logging level).
A few reasons:
- Mux is already widely used in the codebase to refer to a http handler mux.
- Original meaning of Mux was something which sent a chose one of several inputs to
and output. This sends one output to all outputs. Broadcast captures that idea
better.
- Aligns with similar class config.Broadcaster (see #2747)
Implemented via HTTP and websocket. A test is present but this isn't
yet wired into anything.
Eventual purpose of this is to allow a scheduler to watch for new pods.
Or allow replication controller to watch for new items it controlls.
Generally, it'll be good to turn everything possible into a push instead
of a poll.
Alex Robinson