From 6e1debe8cd082044730e05749516c1d0354a6c0c Mon Sep 17 00:00:00 2001 From: jbjohnso Date: Fri, 27 Aug 2010 13:58:40 +0000 Subject: [PATCH] document TODOs in kvm git-svn-id: https://svn.code.sf.net/p/xcat/code/xcat-core/trunk@7280 8638fb3e-16cb-4fca-ae20-7b5d299a9bcd --- xCAT-server/lib/xcat/plugins/kvm.pm | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/xCAT-server/lib/xcat/plugins/kvm.pm b/xCAT-server/lib/xcat/plugins/kvm.pm index 587eecec1..8d3242037 100644 --- a/xCAT-server/lib/xcat/plugins/kvm.pm +++ b/xCAT-server/lib/xcat/plugins/kvm.pm @@ -216,6 +216,7 @@ sub get_filepath_by_url { #at the end of the day, the libvirt storage api gives # incidentally, promote to master will be relatively expensive compared to the converse operation, as expected # will have to verify as it is investigated whether this can successfully cross pools (hope so) # 4) qemu-img was so much more transparent and easy to figure out than this + # additionally, when mastering a powered down node, we should rebase the node to be a cow clone of the master it just spawned } else { my $vol = $poolobj->create_volume("".$desiredname."".getUnits($create,"G",1)."0"); if ($vol) { return $vol->get_path(); } @@ -795,6 +796,13 @@ sub xhrm_satisfy { foreach (@nics) { s/=.*//; #this code cares not about the model of virtual nic $rc |=system("ssh $hyp xHRM bridgeprereq $_"); + #TODO: surprise! there is relatively undocumented libvirt capability for this... + #./tests/interfaceschemadata/ will have to do in lieu of documentation.. + #note that RHEL6 is where that party starts + #of course, they don't have a clean 'migrate from normal interface to bridge' capability + #consequently, would have to have some init script at least pre-bridge it up.. + #even then, may not be able to intelligently modify the bridge remotely, so may still not be feasible for our use.. + #this is sufficiently hard, punting to 2.6 at least.. } return $rc; } @@ -804,6 +812,7 @@ sub makedom { my $xml = shift; my $dom; if (not $xml and $confdata->{kvmnodedata}->{$node} and $confdata->{kvmnodedata}->{$node}->[0] and $confdata->{kvmnodedata}->{$node}->[0]->{xml}) { +#TODO: in this case, build_diskstruct won't be called to do storage validation $xml = $confdata->{kvmnodedata}->{$node}->[0]->{xml}; } elsif (not $xml) { $xml = build_xmldesc($node,$cdloc); @@ -1321,6 +1330,8 @@ sub power { return (1,"Failure satisfying networking and storage requirements on $hyp for $node"); } } +#TODO: here, storage validation is not necessarily performed, consequently, must explicitly do storage validation +#this worked before I started doing the offline xml store because every rpower on tried to rebuild ($dom,$errstr) = makedom($node,$cdloc); if ($errstr) { return (1,$errstr); } } else {