Releases: pinterest/elixometer
Releases · pinterest/elixometer
1.5.0
1.4.1
1.4.0
1.3.0
Added
- Made the histogram
truncate
option a function argument forupdate_histogram/4
(#56) - Made metric name formatting more efficient (#62, #74)
- Support passing extra
subscribe_options
to reporters that accept them (#57, #58) - Support all Elixir time units (#72)
- Support using system environment variables in
:env
configuration (#80) - Support filtering datapoints in subscriptions (#91)
- Support wildcard keys (#97)
- Add typespecs to public Elixometer methods (#107)
- Support configuring the formatter using a module in addition to a function ref (#114)
- Support for bulk subscriptions to get all metrics at once (#134)
Changed
- Elixometer now requires Elixir 1.5 or later
- Lager 3.2.1 or later is now required
- :exometer_core 1.5 or later is now required
Bug fixes
1.3.0 RC1
Added
- Made the histogram
truncate
option a function argument forupdate_histogram/4
(#56) - Made metric name formatting more efficient (#62, #74)
- Support passing extra
subscribe_options
to reporters that accept them (#57, #58) - Support all Elixir time units (#72)
- Support using system environment variables in
:env
configuration (#80) - Support filtering datapoints in subscriptions (#91)
- Support wildcard keys (#97)
- Add typespecs to public Elixometer methods (#107)
- Support configuring the formatter using a module in addition to a function ref (#114)
- Support for bulk subscriptions to get all metrics at once (#134)
Changed
- Elixometer now requires Elixir 1.5 or later
- Lager 3.2.1 or later is now required
- :exometer_core 1.5 or later is now required
Bug fixes
1.2.1
1.2.0
Removed bottlenecks
Elixometer's main gen_server was creating a bottleneck, since it waited for responses from exometer in order to update a stat. This version changes the architecture somewhat such that updating stats is asynchronous and the process that receives the messages has a cap of 1000 messages in its inbox.
It's worth noting that this bottleneck would only occur under extremely high load situations that would send hundreds of thousands of stats per second.