2 MS is typical. There is also an issue with starting stopped MS. For example if you start 1MS and within next 30 seconds start the second MS (first MS is loading all hosts), there is a chance they will come up - but really not functional.
Specifically sync jobs would fail and status updates from cloudstack managed resource may not update. Solution was to stagger the start of the second MS, but its just work around solution. On 11/20/15 2:19 AM, Rafael Weingärtner wrote: > First I thought the problem could be a consequence of over commitment, but > per your answers I do not see problems with over provisioning of resources. > I would check the MS and KVM agent code, but I believe the problem is > number of MS, imagine a singles MS to manage, monitor and orchestrate 250+ > physical servers, networking devices and probably a huge number of VMs. In > the environment that we have in our lab, we use 2 MS to manage a pretty > small environment, and sometimes they get slow. > > On Fri, Nov 20, 2015 at 6:25 AM, Daan Hoogland <daan.hoogl...@gmail.com> > wrote: > >> On Thu, Nov 19, 2015 at 9:31 PM, ilya <ilya.mailing.li...@gmail.com> >> wrote: >> >>> Maybe I need to ping LeaseWeb and ExtremePC folks.. >>> >> >> We have many smaller installations, nothing near those limits. (we being >> LeaseWeb) >> >> >> >> -- >> Daan >> > > >
