Question About behavior on load balanced environments.

[toni-moreno]

Hi @MattHodge I'm using your powerful tools to send windows stats to graphite and we are working on a environment with load balancers in front of several graphite carbon daemons.

I would like to know how Graphite-powershell-Functions works when configured as a service on this environment. ( always assuming TCP connections)

In a failover contingency scenario ( one graphite carbon fails and close all its TCP connections).

• is Graphite-Powershell-Service able to do a reconnection ? in that case which is reconnection frequency until it finally reconnects?

In a recovery scenario (we can recover a carbon server after a crash and load balancers are beginning to distribute load again over all our carbon servers).

• is Graphite-Powershell-Service able to Force a Reconnect (periodically each 3 or 6 hours perhaps) to enabling load redistribution over all carbon servers?

Thank you very much

[MattHodge]

Hey @toni-moreno.

is Graphite-Powershell-Service able to do a reconnection?
in that case which is reconnection frequency until it finally reconnects?

A new connection to the server is created and torn down at every interval defined by your setting for <MetricSendIntervalSeconds> in the configuration file for both TCP and UDP connections. It should reconnect just fine.

is Graphite-Powershell-Service able to Force a Reconnect (periodically each 3 or 6 hours perhaps) to enabling load redistribution over all carbon servers?

As it reconnects every time metrics are sent, this won't be a problem :) Happens every few seconds.