Hi Guillaume, We use our own SPM <http://sematext.com/spm> to monitor our Kafka clusters (producers, brokers, and consumers .... you can actually see all their metrics at [0]) and, while we are currently not using Storm, SPM itself can monitor it (see [1]), which means you can have your Kafka and Storm metrics and alerts side by side, on the same dashboard(s). We do this with other parts of our stack (e.g. Kafka + HBase or Kafka + Elasticsearch) and it works nicely. Oh, and you could still capture other metrics with Coda Hale's Metrics and graph that (see [2]).
[0] https://apps.sematext.com/demo [1] http://blog.sematext.com/2014/01/30/announcement-apache-storm-monitoring-in-spm/ [2] https://sematext.atlassian.net/wiki/display/PUBSPM/Custom+Metrics#CustomMetrics-CodaHale(akaYammer)MetricsReporter Otis -- Monitoring * Alerting * Anomaly Detection * Centralized Log Management Solr & Elasticsearch Support * http://sematext.com/ On Mon, Sep 15, 2014 at 9:53 AM, Guillaume Boesch <boesc...@gmail.com> wrote: > Thank you for your response, I was thinking about using CodaHale's Metrics > for my Trident application but I wanted to be sure that Storm UI and the > REST API were useless in my case. > > 2014-09-13 6:08 GMT+02:00 Cyrille Karmann <cyrillekarm...@gmail.com>: > >> Metrics are a facility designed for Storm, and has no knowledge of >> Trident, which is an abstraction on top of Storm. So messages will appear >> that are used by Trident under the hood. >> To have statistics for Trident, you are a bit on your own. But it is easy >> to use a general metrics library like >> https://dropwizard.github.io/metrics to obtain information from a worker >> via JMX for example. >> >> Cyrille >> > >
