[pve-devel] [PATCH manager 2/7] Broadcast supported CPU flags

[Fabian Grünbichler]

On September 2, 2019 4:27 pm, Stefan Reiter wrote:
> pvestatd will read supported CPU flags once on startup (since these
> never change during runtime, and QEMU updates trigger a service
> restart), then broadcasts them as a key-value pair to the cluster.
> 
> Signed-off-by: Stefan Reiter <s.reiter at proxmox.com>
> ---
> 
> Changes from RFC:
> 
> * Cache joined value
> * Add eval around QEMU query
> * Detect value regression (e.g. "better" value available in cluster kv) and
>   handle accordingly
> 
> 
>  PVE/Service/pvestatd.pm | 28 ++++++++++++++++++++++++++--
>  1 file changed, 26 insertions(+), 2 deletions(-)
> 
> diff --git a/PVE/Service/pvestatd.pm b/PVE/Service/pvestatd.pm
> index e138b2e8..84258daf 100755
> --- a/PVE/Service/pvestatd.pm
> +++ b/PVE/Service/pvestatd.pm
> @@ -47,12 +47,15 @@ my $cmdline = [$0, @ARGV];
>  my %daemon_options = (restart_on_error => 5, stop_wait_time => 5);
>  my $daemon = __PACKAGE__->new('pvestatd', $cmdline, %daemon_options);
>  
> +my $supported_cpuflags;
> +
>  sub init {
>      my ($self) = @_;
>  
>      $opt_debug = $self->{debug};
>  
>      PVE::Cluster::cfs_update();
> +    update_supported_cpuflags();

but this only happens once, not on every status update cycle like all 
the other update_foo subs? if we go the "cache based on 
kvm_user_version" route, then this could actually move to update_status

>  }
>  
>  sub shutdown {
> @@ -72,6 +75,25 @@ sub hup {
>      $restart_request = 1;
>  }
>  
> +sub update_supported_cpuflags {

at least if this is reworked:

1.) get kvm_user_version
2.) check if cached kvm_user_version is different
3.) yes -> query_supported_cpu_flags, cache results + kvm_user_version
4.) yes -> handle regressions
5.) broadcast supported flags

> +    eval {
> +	$supported_cpuflags = join(" ", @{PVE::QemuServer::query_supported_cpu_flags()});
> +    };
> +    warn $@ if $@;
> +
> +    # detect regression
> +    if (!$supported_cpuflags || $supported_cpuflags eq '') {
> +	my $prev_cpuflags = PVE::Cluster::get_node_kv('cpuflags', $nodename)->{$nodename};
> +	if ($prev_cpuflags && $prev_cpuflags ne '') {
> +	    $supported_cpuflags = $prev_cpuflags;
> +	    warn "CPU flag detection failed, using old values\n";
> +	} else {
> +	    $supported_cpuflags = '';
> +	    warn "CPU flag detection failed and no previous values found\n";
> +	}
> +    }
> +}
> +
>  my $generate_rrd_string = sub {
>      my ($data) = @_;
>  
> @@ -91,7 +113,9 @@ sub update_node_status {
>  
>      my $cpuinfo = PVE::ProcFSTools::read_cpuinfo();
>  
> -    my $maxcpu = $cpuinfo->{cpus}; 
> +    my $maxcpu = $cpuinfo->{cpus};
> +
> +    PVE::Cluster::broadcast_node_kv('cpuflags', $supported_cpuflags);

this can then be replaced with a call to update_supported_cpuflags

>  
>      my $subinfo = PVE::INotify::read_file('subscription');
>      my $sublevel = $subinfo->{level} || '';
> @@ -104,7 +128,7 @@ sub update_node_status {
>  	$netin += $netdev->{$dev}->{receive};
>  	$netout += $netdev->{$dev}->{transmit};
>      }
> - 
> +
>      my $meminfo = PVE::ProcFSTools::read_meminfo();
>  
>      my $dinfo = df('/', 1);     # output is bytes
> -- 
> 2.20.1
> 
> 
> _______________________________________________
> pve-devel mailing list
> pve-devel at pve.proxmox.com
> https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
> 
>