Good point David :-) Patch is here: https://reviews.apache.org/r/5655/
Cheers, Hugo -----Original Message----- From: David Nalley [mailto:da...@gnsa.us] Sent: Thursday, June 28, 2012 11:14 PM To: cloudstack-dev@incubator.apache.org Subject: Re: add clouddev commit by Edison Su On Jun 28, 2012, at 11:43 AM, Edison Su <edison...@citrix.com> wrote: > > >> -----Original Message----- >> From: Hugo Trippaers [mailto:htrippa...@schubergphilis.com] >> Sent: Thursday, June 28, 2012 2:51 AM >> To: cloudstack-dev@incubator.apache.org >> Subject: add clouddev commit by Edison Su >> >> Heya, >> >> I was browsing through the clouddev commit and noticed that the >> xen.check.hvm parameter is no hardcoded to false: >> @@ -457,8<https://git-wip-us.apache.org/repos/asf?p=incubator- >> cloudstack.git;a=blob;f=plugins/hypervisors/xen/src/com/cloud/hypervi >> so >> r/xen/discoverer/XcpServerDiscoverer.java;h=8a6c60546d024c3bb9f4cef9a >> 09 b336f07d3ab76;hb=8a6c60546d024c3bb9f4cef9a09b336f07d3ab76#l457> >> +480,7<https://git-wip-us.apache.org/repos/asf?p=incubator- >> cloudstack.git;a=blob;f=plugins/hypervisors/xen/src/com/cloud/hypervi >> so >> r/xen/discoverer/XcpServerDiscoverer.java;h=4fd202beedcfc2f2d7841c30b >> 59 891b1a9362363;hb=4fd202beedcfc2f2d7841c30b59891b1a9362363#l480> @@ >> public class XcpServerDiscoverer extends DiscovererBase implements >> Discoverer, L >> Boolean.parseBoolean(value); >> >> value = _params.get("xen.check.hvm"); >> - _checkHvm = value == null ? true : Boolean.parseBoolean(value); >> - >> + _checkHvm = false; >> _connPool = XenServerConnectionPool.getInstance(); >> >> _agentMgr.registerForHostEvents(this, true, false, true); >> > > Put checkhvm to false, meaning allow admin add a hypervisor host with no hvm > support. > While when creating a VM, there is another check against hvm in > firstfitallocator. > So setting it's to false, will not impact the current functionality. > >> I'm not sure if this is a good idea, the 'old' default is true and >> that makes it a conscious decision by the admin to allow non hvm >> boxes. If we want it like this should we remove the parameter completely? > > Agree, it's better to set xen.check.hvm in configuration xml, other than > hardcode it to false. > I'll fix it later. Hugo, feel free to submit a patch. I am sure Edison would welcome the help. --David
