* [pve-devel] [PATCH qemu-server] fix #5528: override cgroup methods to call systemd via dbus
@ 2024-07-09 9:10 Wolfgang Bumiller
2024-07-24 12:26 ` [pve-devel] applied: " Thomas Lamprecht
0 siblings, 1 reply; 2+ messages in thread
From: Wolfgang Bumiller @ 2024-07-09 9:10 UTC (permalink / raw)
To: pve-devel
Systemd reapplies its known values on reload, so we cannot simply call
into PVE::CGroup. Call systemd's SetUnitProperties method via dbus
instead.
The hotplug and startup code also calculated different values, as one
operated within systemd's value framework (documented in
systemd.resource-control(5)) and one worked with cgroup values
(distinguishing between cgroup v1 and v2 manually).
This is now unified by overriding `change_cpu_quota()` and
`change_cpu_shares()` via `PVE::QemuServer::CGroup` which now takes
systemd-based values and sends those directly via dbus.
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
---
PVE/QemuServer.pm | 4 ++--
PVE/QemuServer/CGroup.pm | 51 +++++++++++++++++++++++++++++++++++++++-
2 files changed, 52 insertions(+), 3 deletions(-)
diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index 88c274d..6e78cff 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -5100,8 +5100,8 @@ sub vmconfig_hotplug_pending {
my $new_cpuunits = PVE::CGroup::clamp_cpu_shares($conf->{pending}->{$opt}); #clamp
$cgroup->change_cpu_shares($new_cpuunits);
} elsif ($opt eq 'cpulimit') {
- my $cpulimit = $conf->{pending}->{$opt} == 0 ? -1 : int($conf->{pending}->{$opt} * 100000);
- $cgroup->change_cpu_quota($cpulimit, 100000);
+ my $cpulimit = $conf->{pending}->{$opt} == 0 ? undef : int($conf->{pending}->{$opt} * 100);
+ $cgroup->change_cpu_quota($cpulimit, undef);
} elsif ($opt eq 'agent') {
vmconfig_update_agent($conf, $opt, $value);
} else {
diff --git a/PVE/QemuServer/CGroup.pm b/PVE/QemuServer/CGroup.pm
index 479991e..882163f 100644
--- a/PVE/QemuServer/CGroup.pm
+++ b/PVE/QemuServer/CGroup.pm
@@ -2,7 +2,10 @@ package PVE::QemuServer::CGroup;
use strict;
use warnings;
-use PVE::CGroup;
+
+use Net::DBus qw(dbus_uint64 dbus_boolean);
+
+use PVE::Systemd;
use base('PVE::CGroup');
sub get_subdir {
@@ -11,4 +14,50 @@ sub get_subdir {
return "qemu.slice/$vmid.scope/";
}
+sub scope {
+ my ($self) = @_;
+ return $self->{vmid} . '.scope';
+}
+
+my sub set_unit_properties : prototype($$) {
+ my ($self, $properties) = @_;
+
+ PVE::Systemd::systemd_call(sub {
+ my ($if, $reactor, $finish_cb) = @_;
+ $if->SetUnitProperties($self->scope(), dbus_boolean(1), $properties);
+ return 1;
+ });
+}
+
+# Update the 'cpulimit' of VM.
+# Note that this is now the systemd API and we expect a value for `CPUQuota` as
+# set on VM startup, rather than cgroup values.
+sub change_cpu_quota {
+ my ($self, $quota, $period) = @_;
+
+ die "period is not controlled for VMs\n" if defined($period);
+
+ $quota = dbus_uint64(defined($quota) ? ($quota * 10_000) : -1);
+ set_unit_properties($self, [ [ CPUQuotaPerSecUSec => $quota ] ]);
+
+ return;
+}
+
+# Update the 'cpuunits' of a VM.
+# Note that this is now the systemd API and we expect a value for `CPUQuota` as
+# set on VM startup, rather than cgroup values.
+sub change_cpu_shares {
+ my ($self, $shares) = @_;
+
+ $shares //= -1;
+
+ if (PVE::CGroup::cgroup_mode() == 2) {
+ set_unit_properties($self, [ [ CPUWeight => dbus_uint64($shares) ] ]);
+ } else {
+ set_unit_properties($self, [ [ CPUShares => dbus_uint64($shares) ] ]);
+ }
+
+ return;
+}
+
1;
--
2.39.2
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
^ permalink raw reply [flat|nested] 2+ messages in thread
* [pve-devel] applied: [PATCH qemu-server] fix #5528: override cgroup methods to call systemd via dbus
2024-07-09 9:10 [pve-devel] [PATCH qemu-server] fix #5528: override cgroup methods to call systemd via dbus Wolfgang Bumiller
@ 2024-07-24 12:26 ` Thomas Lamprecht
0 siblings, 0 replies; 2+ messages in thread
From: Thomas Lamprecht @ 2024-07-24 12:26 UTC (permalink / raw)
To: Proxmox VE development discussion, Wolfgang Bumiller
Am 09/07/2024 um 11:10 schrieb Wolfgang Bumiller:
> Systemd reapplies its known values on reload, so we cannot simply call
> into PVE::CGroup. Call systemd's SetUnitProperties method via dbus
> instead.
>
> The hotplug and startup code also calculated different values, as one
> operated within systemd's value framework (documented in
> systemd.resource-control(5)) and one worked with cgroup values
> (distinguishing between cgroup v1 and v2 manually).
>
> This is now unified by overriding `change_cpu_quota()` and
> `change_cpu_shares()` via `PVE::QemuServer::CGroup` which now takes
> systemd-based values and sends those directly via dbus.
>
> Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
> ---
> PVE/QemuServer.pm | 4 ++--
> PVE/QemuServer/CGroup.pm | 51 +++++++++++++++++++++++++++++++++++++++-
> 2 files changed, 52 insertions(+), 3 deletions(-)
>
>
applied, thanks!
_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2024-07-24 12:26 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-07-09 9:10 [pve-devel] [PATCH qemu-server] fix #5528: override cgroup methods to call systemd via dbus Wolfgang Bumiller
2024-07-24 12:26 ` [pve-devel] applied: " Thomas Lamprecht
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox