This probably no longer compiles. You should use the built-in HTTPOutput. There is even an InfluxDB-specific encoder now.
InfluxDB Output for Mozilla Heka
Example config:
[InfluxDBOutput]
message_matcher = "TRUE"
server = "http://localhost:8086"
database = "db"
username = "user"
password = "pass"
# InfluxDB connection timeout, in seconds
response_header_timeout = 30
# The default if series in the field_map is missing
series = "history"
# Use the Heka message Timestamp as the InfluxDB time; on by default
use_heka_timestamp = true
# Flush points to InfluxDB at least this often, in milliseconds
flush_interval = 1000
# Buffer no more than this many points before flushing to InfluxDB
flush_count = 10
[InfluxDBOutput.field_map]
# Map InfluxDB field names to Heka message field names
payload = "Payload"
custom_field = "CustomField"
# Special case for dynamic series name - falls back to series above
series = "Type"