Logging latency and throughput

Tor Bonde

unread,

Dec 6, 2013, 9:00:23 AM12/6/13

to storm...@googlegroups.com

If I want to log certain metrics (like latency and throughput), what would be the best way of doing that?

I have thought of using hooks, but I'm not sure whether that's a bad idea. When I run a topology with several workers, will an assigned hook be instantiated multiple times as well? I'm worried that this could screw up my metrics.

Andrew Neilson

unread,

Dec 7, 2013, 3:41:47 PM12/7/13

to storm...@googlegroups.com

I've found this blog post by Patricio Echague helpful. It might give you some ideas too: http://blog.relateiq.com/monitoring-storm/

On Fri, Dec 6, 2013 at 6:00 AM, Tor Bonde <cimr...@gmail.com> wrote:

If I want to log certain metrics (like latency and throughput), what would be the best way of doing that?

I have thought of using hooks, but I'm not sure whether that's a bad idea. When I run a topology with several workers, will an assigned hook be instantiated multiple times as well? I'm worried that this could screw up my metrics.

--
You received this message because you are subscribed to the Google Groups "storm-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to storm-user+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.