These start a Pipe by feeding it events on a line-by-line basis.

Unless otherwise specified, the specified arguments to input can contain {{var}} expansions, where var can be a variable or simple chain of variables separated by '.' from the Context

By default, you can assume that the output will be in JSON format. For an input that already generates JSON, say raw: true, or json: true. The latter is deprecated because what we mean is that the input passes through without being quoted like so {"_raw":"line"}.

Many of these inputs support retrying:

        count: 3 # number of times to retry
        pause: 2s # pause between retries
        forever: false # instead of count, just keep rerying

Inputs either block until data arrives, or are scheduled.

Inputs that can be scheduled (like http-poll and redis hash value lookup) can be used as actions. In this case event field expansions are available, and any existing data is preserved by merging whatever is received.

files (input)Read from multiple files, in order of creation
redis (input)Read from Redis in-memory key-value store
amqp (input)Read from AMQP queues
tcp (input)Listen for incoming TCP connections (or write to existing server)
udp (input)Listen for incoming UDP connections
http-poll (input)Run HTTP queries (GET and POST)
http-server (input)Run HTTP server
exec (input)Execute arbitrary commands