Chiradeep, For autoscale in the current model the monitoring + trigger happens from the load balancer device. It then issues CS API calls to deploy/destroy vms. How can we make it work?
Thanks, Ram > -----Original Message----- > From: Chiradeep Vittal [mailto:chiradeep.vit...@citrix.com] > Sent: 11 July 2012 23:14 > To: CloudStack DeveloperList > Subject: Re: Network element's access to CS mgmt server....? > > Certainly user VMs cannot. But system vms can access the management > server, typically on port 8250. > One option might be to poll the VPX (depending on how many there are) > for > autoscaling triggers from the MS. > > > On 7/11/12 6:18 AM, "Ram Ganesh" <ram.gan...@citrix.com> wrote: > > >Alex, > > > >> You need to start the vpx with a nic in the management network. In > >> CloudStack code, SSVM (Secondary Storage Virtual Machine) and CPVM > >> (Console Proxy Virtual Machine) both have examples of this. They > both > >> have agents that connect back to CS Management server over the > >> management network. > > > >Yes we can add a nic(vif) to a vpx in the management network and > ensure > >reach ability to the CS mgmt server. But how about the CS mgmt ports > such > >as 8080/443 - they may be blocked by intermediate firewall/VR? Also > >please note unlike SSVM from vpx we may be able to issue a finite ( > >restricted) set of CS API calls so we may not be able to issue > firewall > >rule related API calls. Also the load balancer devices can be placed > into > >network from outside the context of CloudStack and later the device > can > >be added into CloudStack using addNetscalerLoadbalancer()API. > > > >So to put it another way - Can the VMs spread across various guest > >networks reach out to the CS mgmt server using a same IP ? > > > > > >Thanks, > >Ram > >
