graphite-dev team mailing list archive

Thread
Date

[Question #227324]: metric order differ when using wildcard in cluster

To: graphite-dev@xxxxxxxxxxxxxxxxxxx
From: "R. Toma" <question227324@xxxxxxxxxxxxxxxxxxxxx>
Date: Tue, 23 Apr 2013 20:51:16 -0000
Reply-to: question227324@xxxxxxxxxxxxxxxxxxxxx
Sender: bounces@xxxxxxxxxxxxx

New question #227324 on Graphite:
https://answers.launchpad.net/graphite/+question/227324

Hi all,

I have two nodes. Both run 1x relay, 2x cache and 1x webapp. With a VIP in front of the webapps and relays.

When requesting "target=aliasByNode(carbon.agents.*.cpuUsage,2)" directly on the nodes the order of the target depend on the webapp/node used.

The wildcard resolves into this list:
* carbon.agents.node1-cache1.cpuUsage
* carbon.agents.node1-cache2.cpuUsage
* carbon.agents.node2-cache1.cpuUsage
* carbon.agents.node2-cache2.cpuUsage

Requesting on node1 the order is: node1 agents first, followed by node2 agents.
Requesting on node2 the order is: node2 agents first, followed by node1 agents.

So when requesting this graph multiple times via a VIP I get different graphs.

Is this normal behaviour, or do I have a misconfiguration?

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