v0.0.21

upd: only update metric counter for tracking metrics for non-sub output
add: latency metrics for metric processing
add: submit timestamp based on metric timestamp
add: circonus debugging specific cgm check
fix: cache dir check
upd: testing cache directory
add: new memcached metrics
add: apiclient max retries, min delay, max delay settings to address duplicate checks
upd: dependency (cgm, go-apiclient)
add: batch processing pool
add: pool_size config option
add: check config caching
add: cache_configs and cache_dir options
add: dump_cgm_metrics option for debugging
add: sub_output option to tell 2-n instances of circonus output to not initialize default checks
add: dynamic_submit option to let cgm submit metrics on its internal cadence
add: dynamic_interval option to set the cgm interval
fix: overwrite existing numeric rather than add - duplicate metrics in same batch
add: back alias tag for multiple instances
add: comment on limitnofile for 'too many open files' errors with large number of inputs
fix: load defaults once per agent run, not on config reload
upd: use instance_id for alias on inputs if alias not set (for logging)
upd: lint config
upd: add input_id to all inputs in sample config
upd: debug messages
upd: remove commented code
upd: switch version parser
upd: snmp dynamic textual conversion for tags
add: snmp automatic tag lookups
add: additional counters to the windows_perf_counters input plugin for Windows host monitoring service dashboard

Docker images