List view
- No due date•8/8 issues closed
- No due date•16/16 issues closed
We've learned a lot about cernan's internal model and how it ought to be changed to improve aggregation performance _and_ support sources/sinks with distinct ideas about things. This milestone pays down a lot of domain mismatch and _should_ simplify cernan's internal model.
No due date•9/9 issues closedIn this milestone cernan users will be able to use scripts as filters between sources and sinks to do runtime analysis of their telemetry and logs. Cernan will also be improved in its operation, with logging and such.
No due date•48/48 issues closedThe primary innovation of this release will be the use of disk-based queues instead of in-memory rust channels. The primary killer of heka's performance was the use of in-memory channels as the approach does not allow for back pressure in the event of surging load. Disk queues open up the possibility of arbitrary topology between sources and sinks, much needed.
No due date•24/24 issues closed- No due date•1/1 issues closed
- No due date•31/31 issues closed
The fundamental storage of cernan has been changed and can now accept graphite points. This release significantly reduces copying and is better set for high performance log manipulation.
No due date•16/16 issues closedFirst production worthy cut of cernan. It is able to multiplex over librato and wavefront, eating etsy/statsd's cake.
No due date•3/3 issues closed