I may be wrong in this, but I believe that you ought to be able to set the VMs up so that they don't automatically boot up when the system boots. You can then write a manifest file that boots your VMs up, one at a time, in any order that you want, with any delay between them that you want. You can even do something to check the status of each boot and, based on the status, determine follow up actions. For example, as each VM boots, have it write to a file that's accessible to the global zone. Your manifest will then check the file to determine whether the boot succeeded.
This is just off the top of my head, and there may be better solutions easily available. This is a easy and cheap (free!) solution that might work for you. Regards, John On Dec 08, 2015, at 08:52 AM, George Mamalakis <[email protected]> wrote: Hi everybody, We are looking for a virtualisation solution and during my search I came across SmartOS. Our project will not be large, it will be initially comprised of 4-5 48core machines and we will need to be able to add physical machines no more than once or twice a year. We need to be able to have network separation among some of our VMs (so we need virtual network support, 802.1q vlan support, etc), we want to have a (web) interface to manage VMs and we wish our solution of choice to allow for high availability, in the sense that a VM will need to be able to maintain its state if one or more of the underlying physical devices have crashed, depending on our configuration. I've read that SmartOS supports live migration and high availability through Quemu using SmartDataCenter, and I also read that FiFo accomplishes something similar using LeoFS. As far as LeoFS is concerned, its documentation mentions that the consistency level is responsible of the redundancy that is established, but this configuration is performed only once, during the design phase. This, to my understanding, is a somewhat limiting factor, since even in our setup, which will not change more than twice a year (by adding new physical nodes), we cannot be certain from day one how much our infrastructure will grow in order to plan ahead. And even if we did, I don't know how much the additional write replicas that will have to be assigned during the design phase -for example- will affect the overall performance. On the other hand, I am not sure how SDC's high availability solution works, and hence I decided to ask this list how high availability can be achieved in a SmartOS-based virtualisation environment. Some more requirements I was given by my team involve configurable VM boot order and docker, but I explained them that the latter is not mandatory. Thank you for your time in advance, George Mamalakis -- George Mamalakis IT and Security Officer, Electrical and Computer Engineer (Aristotle Univ. of Thessaloniki), PhD (Aristotle Univ. of Thessaloniki), MSc (Imperial College of London) School of Electrical and Computer Engineering Aristotle University of Thessaloniki phone number : +30 (2310) 994379 ------------------------------------------- smartos-discuss Archives: https://www.listbox.com/member/archive/184463/=now RSS Feed: https://www.listbox.com/member/archive/rss/184463/25769125-55cfbc00 Modify Your Subscription: https://www.listbox.com/member/?member_id=25769125&id_secret=25769125-7688e9fb Powered by Listbox: http://www.listbox.com
