graphite-dev team mailing list archive

Thread
Date

[Question #240882]: Relay metrics received vs metrics sent

To: graphite-dev@xxxxxxxxxxxxxxxxxxx
From: Franklin Angulo <question240882@xxxxxxxxxxxxxxxxxxxxx>
Date: Mon, 16 Dec 2013 19:41:23 -0000
Reply-to: question240882@xxxxxxxxxxxxxxxxxxxxx
Sender: bounces@xxxxxxxxxxxxx

New question #240882 on Graphite:
https://answers.launchpad.net/graphite/+question/240882

I'm trying to understand the Carbon-generated metrics for the relays. I have a setup with multiple relays receiving metrics from a load balancer. Using aggregated-consistent-hashing the relays then send the metrics to multiple aggregators.

I ran a test in which I keep sending additional metrics to the load balancer. At one point the number of metrics received at the relays exceeds the number of metrics sent to the aggregators. See the image below:

http://tinypic.com/r/15ydyk5/5

What does this mean? 

- Are metrics getting dropped by the relays before they are sent to the aggregators?
- Or are the relays sending batches of metrics thus reducing the *sent* metric reported?

-- 
You received this question notification because you are a member of
graphite-dev, which is an answer contact for Graphite.