Doubt about lots of persistent connections

[Dani Gutiérrez Porset]

Hi community,

in a testing PE installation I see each agent maintains a persistent collection to ActiveMQ process in master server, to be used by mcollective in a push (vs pull) scheme.

As it's not so sure that one master is sending lots of petitions to agents, which are the advantages of this schema instead of non-persistent connections, specially for the case that there are lots of agents? (well, maybe this is a question more for ActiveMQ than Puppet, but...)

Could anyone tell about the hardware (mainly RAM) for the master server (with ActiveMQ) supporting 500 or 1000 Puppet agents?

  Thanks a lot in advance

[Dick Davies]

Those connections don't take much in the way of resources on the ActiveMQ server
(you might need to up the number of open files the activemq process is
able to access).

I'm not sure what you mean about non-persistent connections; are you
talking about
agents periodically connecting to check? i'd expect the continual
connection setup/teardown
to put more load on the MQ server that 1000 or so idle connections
(with maybe the occasional
keepalive). STOMP is pretty lightweight.

> --
> You received this message because you are subscribed to the Google Groups
> "Puppet Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to puppet-users...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/puppet-users/60dea08d-fc3f-4da1-aae2-06ef528fa07a%40googlegroups.com.
> For more options, visit https://groups.google.com/groups/opt_out.