[pve-devel] [PATCH v2 qemu-server 5/5] hotplug_pending: allow partial fast plugging

[Oguz Bektas]

On Thu, Mar 12, 2020 at 02:18:07PM +0100, Thomas Lamprecht wrote:
> On 3/12/20 2:11 PM, Oguz Bektas wrote:
> > On Tue, Mar 10, 2020 at 11:25:23AM +0100, Thomas Lamprecht wrote:
> >> On 2/19/20 5:07 PM, Oguz Bektas wrote:
> >>> adds a loop after the main fastplug loop, to check if any of the options
> >>> are partially fast pluggable.
> >>>
> >>> these are defined in $partial_fast_plug_option
> >>>
> >>> Signed-off-by: Oguz Bektas <o.bektas at proxmox.com>
> >>> ---
> >>>
> >>> v1->v2:
> >>> * set $changes according to partial_fast_plug result as well as fast_plug result
> >>> * do cleanup_pending before writing fastplug changes
> >>>
> >>>
> >>>
> >>>  PVE/QemuConfig.pm |  7 +++++++
> >>>  PVE/QemuServer.pm | 19 +++++++++++++++++++
> >>>  2 files changed, 26 insertions(+)
> >>>
> >>> diff --git a/PVE/QemuConfig.pm b/PVE/QemuConfig.pm
> >>> index 1ba728a..d1727b2 100644
> >>> --- a/PVE/QemuConfig.pm
> >>> +++ b/PVE/QemuConfig.pm
> >>> @@ -399,6 +399,13 @@ sub __snapshot_foreach_volume {
> >>>  
> >>>      PVE::QemuServer::foreach_drive($conf, $func);
> >>>  }
> >>> +
> >>> +sub get_partial_fast_plug_option {
> >>> +    my ($class) = @_;
> >>> +
> >>> +    return $PVE::QemuServer::partial_fast_plug_option;
> >>> +}
> >>> +
> >>>  # END implemented abstract methods from PVE::AbstractConfig
> >>>  
> >>>  1;
> >>> diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
> >>> index 44d0dee..8a689a0 100644
> >>> --- a/PVE/QemuServer.pm
> >>> +++ b/PVE/QemuServer.pm
> >>> @@ -4732,6 +4732,18 @@ my $fast_plug_option = {
> >>>      'tags' => 1,
> >>>  };
> >>>  
> >>> +# name of opt
> >>> +# -> fmt -> format variable
> >>> +# -> properties -> fastpluggable options hash
> >>> +our $partial_fast_plug_option = {
> >>> +    agent => {
> >>> +	fmt => $agent_fmt,
> >>> +	properties => {
> >>> +	    fstrim_cloned_disks => 1
> >>> +	},
> >>> +    },
> >>> +};
> >> this belongs solely in the get_partial_fast_plug_option method, I do not want
> >> to tighten the cyclic use of both modules and it's not required here.
> > when we move the hash to QemuConfig it complains about $agent_fmt which
> > resides in QemuServer (as well as other formats which could be used
> > later on for partial fastplugging). that was the initial reason i added
> > it there because i didn't want to move the formats. should we change the
> > definitions from 'my' to 'our' to use them in QemuConfig? or how to
> > handle this better?
> > 
> 
> Ah yeah, in the longer run the schemas should move to QemuConfig in the longer run,
> IIRC some discussions. 

starting to work on QemuSchema rebase then, since it seems to be the
better option in the long run.