sending metric data to librato as a gauge

[Brice Burgess]

Hi and thanks for l2met! :)

Admittedly I haven't delved in yet; but had a quick ? regarding the ability to limit the metrics sent to librato, and to potentially send metric data as a gauge.

From what I see the librato outlet submits collected metrics to librato as a counter; creating a metric for each calculated aggregate [e.g. my_metric.max = max_val, my_metric.min = min_val, etc. etc. ].

https://github.com/ryandotsmith/l2met/blob/master/outlet/librato_outlet.go#L109

I wanted to know if I'm using the right tool for the job. We have a distributed processing application that works on documents "within a queue". I'd like the worker application to;

 1. output when it completes a task [I believe the built in behavior of l2met is great for this! seems to be a use-case for a "counter"]
 2. [ocassionally?] output the number of tasks left in the queue -- which I believe is a use-case for a "gauge".


Are gauges supported?

We'd also like to limit metric streams to librato for efficiency and $$ sake ;)

Thanks,

~ Brice

[Ryan Smith]

Hey! Sorry for the late reply here. My google groups settings were misconfigured.

Thanks for considering l2met ! :)

I think your use case is definitely within the scope of l2met. One of the things I commonly do is log each stage of the data pipeline. You also might want to track the latency on these operations. E.g.

measure.queue.enqueue=4ms

measure.queue.dequeue=10ms

With these statements you can monitor the performance of your queueing operations and keep track of how many jobs your are processing. Another idea is to have a reporter thread that measures the backlog on your queue and emits the number every second. For the latter suggestion, the current state of l2met will emmitt metrics that are not relevant. However, we are working to address that now. Checkout this issue: https://github.com/ryandotsmith/l2met/issues/69