<html>
<head>
<style><!--
.hmmessage P
{
margin:0px;
padding:0px
}
body.hmmessage
{
font-size: 12pt;
font-family:Calibri
}
--></style></head>
<body class='hmmessage'><div dir='ltr'><br><br><div>> To: rbowen@redhat.com; rdo-list@redhat.com<br>> From: pmyers@redhat.com<br>> Date: Fri, 18 Sep 2015 15:49:39 -0400<br>> Subject: Re: [Rdo-list] RDO-Manager "quickstart"<br>> <br>> On 09/18/2015 03:35 PM, Rich Bowen wrote:<br>> > One of the goals of the RDO Packstack Quickstart is to give people a<br>> > successful and easy first-time experience deploying OpenStack, even if<br>> > what they're left with (an --allinone deployment) might not be, strictly<br>> > speaking, *useful* for much.<br>> > <br>> > Today on IRC I asked if we might possibly work towards a similar<br>> > quickstart for RDO-Manager, where we make a bunch of assumptions and<br>> > automate whatever parts of it we can, and end up with a "do these three<br>> > steps" kind of thing, like the Packstack quickstart.<br>> > <br>> > I've included the transcript of the conversation below, but since IRC<br>> > transcripts can be confusing after the fact, to summarize, slagle opined<br>> > that it might be feasible to have two paths - the full-featured path<br>> > that we currently have, but also something as I described above for your<br>> > first time.<br>> > <br>> > I wanted to toss this out here for a larger audience to see whether this<br>> > seems like a reasonable goal to pursue?<br>> <br>> +1<br>> <br>> I think it's critical to have something that's easy to work for _very_<br>> constrained use cases. But I also agree with the below sentiments that<br>> we need to properly document and enable folks to do more complex<br>> deployments after they've had their first success with the minimal<br>> deployment option<br>> <br>> One thing to consider in all of this is... what is the minimum<br>> deployment footprint? I think we have to assume virtual, since most<br>> folks won't have a lab with 6 nodes sitting around.<br>> <br>> A few options:<br>> <br>> a Undercloud on one VM, single overcloud controller on another VM,<br>> single compute node on another VM (using nested virt, or just plain<br>> emulation)<br>> <br>> b 2nd variation on the above would be to run the 3 node controller HA<br>> setup, which means 1 undercloud, 3 overcloud controllers + 1 compute<br>> <br>> The question is... what is the minimum amount of RAM that you can run an<br>> overcloud controller with? 4GB? Or can that be squeezed to 2 or 3GB<br>> just for playing around purposes?<br>> <br>> What is the minimum amount of RAM you need for the undercloud node?<br>> <br>> If 4GB per VM, then a) maybe can be done on a 16GB system, while b)<br>> needs 32GB<br><br>32 GB RAM is not actually a big problem, but test "b" will require 8 CORE CPU at least like <br>Intel® Xeon® Processor E5-2690 (costs about $2000) and corresponding board, which require <br>business environment and would fail on desktop CPU. Even expensive i7 ( Haswell Kernel) top line<br>models won't provide ability to test "b", only "a" due to 4 CORES limitation even with HT enabled.<br>Please, correct me if I am wrong about that.<br><br>Boris<br><br>> <br>> If we could squeeze controller and undercloud nodes into 3GB each, then<br>> it might be possible to run b) on a 16GB machine, opening up<br>> experimentation with RDO Manager in a real HA configuration to lots more<br>> people<br>> <br>> Perry<br>> <br>> _______________________________________________<br>> Rdo-list mailing list<br>> Rdo-list@redhat.com<br>> https://www.redhat.com/mailman/listinfo/rdo-list<br>> <br>> To unsubscribe: rdo-list-unsubscribe@redhat.com<br></div> </div></body>
</html>