* [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA
@ 2026-11-10 10:45 Jakob Klocker
2026-11-10 10:45 ` [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback Jakob Klocker
` (3 more replies)
0 siblings, 4 replies; 10+ messages in thread
From: Jakob Klocker @ 2026-11-10 10:45 UTC (permalink / raw)
To: pve-devel
This series adds a new agent option 'sync-time-on-resume' to
automatically synchronize the guest clock via the QEMU Guest Agent
after operations that can leave the guest time stale. The option is
disabled by default and only takes effect when the QEMU Guest Agent is
enabled.
When a VM resumes with a restored RAM state (waking from hibernation or
rolling back to a snapshot that includes RAM), the guest clock
continues from the point the state was saved and no longer matches
wall-clock time. Skew also appears whenever the guest is briefly
frozen and resumed - after creating a snapshot, an
ordinary pause/resume, the resume step of a live migration, or a
vzdump in suspend mode.
When enabled, the option triggers a guest-set-time call:
* after resuming from hibernation (suspend-to-disk)
* after rolling back to a snapshot that includes RAM
* after a live migration
* after an ordinary pause/resume
* after taking a snapshot
* after a vzdump backup in suspend mode
The start, resume and rollback API endpoints also accept a
'sync-time-on-resume' parameter that overrides the configured value for
a single operation.
The option is exposed in the VM configuration GUI
so it can be toggled per guest.
Changes since v5:
* drop the migration version check (covered by the strict config check
series and the same-version cluster requirement)
* vzdump: sync the guest clock after resuming from a suspend mode backup;
snapshot mode doesn't pause the vCPUs and stop mode boots the guest
fresh, so neither needs a sync
* vm_resume: skip the sync when resuming from 'prelaunch' (e.g. a VM
started paused for a backup) or after a reset, the guest boots fresh in
both cases
* guest_set_time: return whether the time was actually set
* rebased on current master
Changes since v4:
* rebased on current master
* no functional changes
Link: https://bugzilla.proxmox.com/show_bug.cgi?id=5032
pve-guest-common:
Jakob Klocker (1):
AbstractConfig: allow passing options to snapshot_rollback
src/PVE/AbstractConfig.pm | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
pve-manager:
Jakob Klocker (1):
fix #5032: ui: qemu agent: add sync-time-on-resume option
www/manager6/form/AgentFeatureSelector.js | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
qemu-server:
Jakob Klocker (1):
fix #5032: agent: sync guest time via QGA when the clock falls behind
src/PVE/API2/Qemu.pm | 75 ++++++++++++++++++++++++++++++++--
src/PVE/CLI/qm.pm | 2 +-
src/PVE/QemuConfig.pm | 12 +++++-
src/PVE/QemuMigrate.pm | 3 ++
src/PVE/QemuServer.pm | 18 +++++++-
src/PVE/QemuServer/Agent.pm | 64 +++++++++++++++++++++++++++++
src/PVE/QemuServer/RunState.pm | 15 ++++++-
src/PVE/VZDump/QemuServer.pm | 25 +++++++++++-
8 files changed, 203 insertions(+), 11 deletions(-)
Summary over all repositories:
10 files changed, 224 insertions(+), 13 deletions(-)
--
Generated by murpp 0.12.0
^ permalink raw reply [flat|nested] 10+ messages in thread* [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback 2026-11-10 10:45 [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker @ 2026-11-10 10:45 ` Jakob Klocker 2026-11-13 14:10 ` Fiona Ebner 2026-11-10 10:46 ` [PATCH pve-manager v6 2/3] fix #5032: ui: qemu agent: add sync-time-on-resume option Jakob Klocker ` (2 subsequent siblings) 3 siblings, 1 reply; 10+ messages in thread From: Jakob Klocker @ 2026-11-10 10:45 UTC (permalink / raw) To: pve-devel Thread an optional $opts hashref through snapshot_rollback to __snapshot_rollback_vm_start, so callers can pass per-rollback options down to the VM start. Used by qemu-server to let the API override the guest time sync for a single rollback; other options can be added without further signature changes. Signed-off-by: Jakob Klocker <j.klocker@proxmox.com> --- src/PVE/AbstractConfig.pm | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/PVE/AbstractConfig.pm b/src/PVE/AbstractConfig.pm index 7bcae19..76ea97a 100644 --- a/src/PVE/AbstractConfig.pm +++ b/src/PVE/AbstractConfig.pm @@ -1125,7 +1125,8 @@ my $rollback_remove_replication_snapshots = sub { # Rolls back to a given snapshot. sub snapshot_rollback { - my ($class, $vmid, $snapname) = @_; + my ($class, $vmid, $snapname, $opts) = @_; + $opts //= {}; my $prepare = 1; @@ -1197,7 +1198,7 @@ sub snapshot_rollback { $class->write_config($vmid, $conf); if (!$prepare && $snap->{vmstate}) { - $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data); + $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data, $opts); } }; -- 2.47.3 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback 2026-11-10 10:45 ` [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback Jakob Klocker @ 2026-11-13 14:10 ` Fiona Ebner 2026-11-16 12:23 ` Jakob Klocker 0 siblings, 1 reply; 10+ messages in thread From: Fiona Ebner @ 2026-11-13 14:10 UTC (permalink / raw) To: Jakob Klocker, pve-devel Am 10.11.26 um 11:48 AM schrieb Jakob Klocker: > Thread an optional $opts hashref through snapshot_rollback to > __snapshot_rollback_vm_start, so callers can pass per-rollback options > down to the VM start. Used by qemu-server to let the API override the > guest time sync for a single rollback; other options can be added > without further signature changes. > > Signed-off-by: Jakob Klocker <j.klocker@proxmox.com> > --- > src/PVE/AbstractConfig.pm | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > > diff --git a/src/PVE/AbstractConfig.pm b/src/PVE/AbstractConfig.pm > index 7bcae19..76ea97a 100644 > --- a/src/PVE/AbstractConfig.pm > +++ b/src/PVE/AbstractConfig.pm > @@ -1125,7 +1125,8 @@ my $rollback_remove_replication_snapshots = sub { > > # Rolls back to a given snapshot. > sub snapshot_rollback { > - my ($class, $vmid, $snapname) = @_; > + my ($class, $vmid, $snapname, $opts) = @_; Nit: maybe $vm_start_opts? > + $opts //= {}; > > my $prepare = 1; > > @@ -1197,7 +1198,7 @@ sub snapshot_rollback { > $class->write_config($vmid, $conf); > > if (!$prepare && $snap->{vmstate}) { > - $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data); > + $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data, $opts); > } > }; > ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback 2026-11-13 14:10 ` Fiona Ebner @ 2026-11-16 12:23 ` Jakob Klocker 0 siblings, 0 replies; 10+ messages in thread From: Jakob Klocker @ 2026-11-16 12:23 UTC (permalink / raw) To: Fiona Ebner, pve-devel On Fri Nov 13, 2026 at 3:10 PM CET, Fiona Ebner wrote: > Am 10.11.26 um 11:48 AM schrieb Jakob Klocker: >> Thread an optional $opts hashref through snapshot_rollback to >> __snapshot_rollback_vm_start, so callers can pass per-rollback options >> down to the VM start. Used by qemu-server to let the API override the >> guest time sync for a single rollback; other options can be added >> without further signature changes. >> >> Signed-off-by: Jakob Klocker <j.klocker@proxmox.com> >> --- >> src/PVE/AbstractConfig.pm | 5 +++-- >> 1 file changed, 3 insertions(+), 2 deletions(-) >> >> diff --git a/src/PVE/AbstractConfig.pm b/src/PVE/AbstractConfig.pm >> index 7bcae19..76ea97a 100644 >> --- a/src/PVE/AbstractConfig.pm >> +++ b/src/PVE/AbstractConfig.pm >> @@ -1125,7 +1125,8 @@ my $rollback_remove_replication_snapshots = sub { >> >> # Rolls back to a given snapshot. >> sub snapshot_rollback { >> - my ($class, $vmid, $snapname) = @_; >> + my ($class, $vmid, $snapname, $opts) = @_; > > Nit: maybe $vm_start_opts? ack, will rename in v7 > >> + $opts //= {}; >> >> my $prepare = 1; >> >> @@ -1197,7 +1198,7 @@ sub snapshot_rollback { >> $class->write_config($vmid, $conf); >> >> if (!$prepare && $snap->{vmstate}) { >> - $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data); >> + $class->__snapshot_rollback_vm_start($vmid, $snap->{vmstate}, $data, $opts); >> } >> }; >> ^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH pve-manager v6 2/3] fix #5032: ui: qemu agent: add sync-time-on-resume option 2026-11-10 10:45 [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker 2026-11-10 10:45 ` [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback Jakob Klocker @ 2026-11-10 10:46 ` Jakob Klocker 2026-11-10 10:46 ` [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind Jakob Klocker 2026-11-17 12:41 ` superseded: [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker 3 siblings, 0 replies; 10+ messages in thread From: Jakob Klocker @ 2026-11-10 10:46 UTC (permalink / raw) To: pve-devel Expose the new agent configuration option `sync-time-on-resume` in the VM configuration GUI. This allows enabling/disabling automatic guest clock synchronization after clock-stalling operations. Signed-off-by: Jakob Klocker <j.klocker@proxmox.com> --- www/manager6/form/AgentFeatureSelector.js | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/www/manager6/form/AgentFeatureSelector.js b/www/manager6/form/AgentFeatureSelector.js index 3cae4194..cd668583 100644 --- a/www/manager6/form/AgentFeatureSelector.js +++ b/www/manager6/form/AgentFeatureSelector.js @@ -45,6 +45,20 @@ Ext.define('PVE.form.AgentFeatureSelector', { hidden: '{freeze_fs.checked}', }, }, + { + xtype: 'proxmoxcheckbox', + boxLabel: gettext( + "Synchronize guest clock with the host's after resuming from a previous state", + ), + name: 'sync-time-on-resume', + reference: 'sync_time_on_resume', + bind: { + disabled: '{!enabled.checked}', + }, + disabled: true, + uncheckedValue: '0', + defaultValue: '0', + }, { xtype: 'displayfield', userCls: 'pmx-hint', @@ -75,6 +89,10 @@ Ext.define('PVE.form.AgentFeatureSelector', { delete values['freeze-fs']; } + if (!PVE.Parser.parseBoolean(values['sync-time-on-resume'])) { + delete values['sync-time-on-resume']; + } + const agentstr = PVE.Parser.printPropertyString(values, 'enabled'); return { agent: agentstr }; }, -- 2.47.3 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind 2026-11-10 10:45 [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker 2026-11-10 10:45 ` [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback Jakob Klocker 2026-11-10 10:46 ` [PATCH pve-manager v6 2/3] fix #5032: ui: qemu agent: add sync-time-on-resume option Jakob Klocker @ 2026-11-10 10:46 ` Jakob Klocker 2026-11-13 14:10 ` Fiona Ebner 2026-11-17 12:41 ` superseded: [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker 3 siblings, 1 reply; 10+ messages in thread From: Jakob Klocker @ 2026-11-10 10:46 UTC (permalink / raw) To: pve-devel Add a 'sync-time-on-resume' agent config property (default off) that, when enabled, issues the guest-set-time QGA command to set the guest clock to the host's current time after operations that leave the clock behind: - resuming from hibernation (suspend-to-disk) - resuming from pause/suspend - rolling back to a snapshot that includes RAM - live migration - taking a snapshot, during which the guest is briefly frozen - vzdump backup in suspend mode The start, resume and rollback API endpoints gain a 'sync-time-on-resume' parameter (e.g. 'qm start <vmid> --sync-time-on-resume 0') that overrides the configured agent value for a single operation, forcing the sync on or off regardless of config. guest_set_time prints a confirmation line to stdout on success. During live migration the resume step runs over the migration tunnel, which also uses stdout, so an unconditional print corrupts the tunnel protocol. vm_resume therefore takes a 'quiet' parameter to suppress the print on the migration path; the confirmation is logged by the migration code instead. For vzdump, only suspend mode needs a sync: snapshot mode never pauses the vCPUs, and in stop mode the guest boots fresh. For the same reason, vm_resume skips the sync when resuming from 'prelaunch' (e.g. a VM started paused for a backup) or after a reset of a VM that shut down during a backup. Signed-off-by: Jakob Klocker <j.klocker@proxmox.com> --- src/PVE/API2/Qemu.pm | 75 ++++++++++++++++++++++++++++++++-- src/PVE/CLI/qm.pm | 2 +- src/PVE/QemuConfig.pm | 12 +++++- src/PVE/QemuMigrate.pm | 3 ++ src/PVE/QemuServer.pm | 18 +++++++- src/PVE/QemuServer/Agent.pm | 64 +++++++++++++++++++++++++++++ src/PVE/QemuServer/RunState.pm | 15 ++++++- src/PVE/VZDump/QemuServer.pm | 25 +++++++++++- 8 files changed, 203 insertions(+), 11 deletions(-) diff --git a/src/PVE/API2/Qemu.pm b/src/PVE/API2/Qemu.pm index 7830d09b..df2076a6 100644 --- a/src/PVE/API2/Qemu.pm +++ b/src/PVE/API2/Qemu.pm @@ -3599,6 +3599,24 @@ __PACKAGE__->register_method({ description => 'Whether to migrate conntrack entries for running VMs.', }, 'nets-host-mtu' => get_standard_option('pve-qm-nets-host-mtu'), + 'sync-time-on-resume' => { + type => 'boolean', + description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this start only. Controls whether the guest's clock is" + . " synchronized after the RAM state is restored. The guest agent must be" + . " enabled.", + verbose_description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this start only. When enabled, the 'guest-set-time' QEMU guest" + . " agent command is issued to synchronize the guest's clock to the host's current" + . " time after the RAM state is restored, which happens here when starting a VM" + . " that was hibernated (suspended to disk). Without synchronization the guest's" + . " clock may still reflect the time from when the VM was hibernated. The time is" + . " only synchronized when the QEMU Guest Agent option is enabled in the guest's" + . " configuration and the agent is running inside of the guest.", + optional => 1, + }, }, }, returns => { @@ -3614,6 +3632,7 @@ __PACKAGE__->register_method({ my $vmid = extract_param($param, 'vmid'); my $timeout = extract_param($param, 'timeout'); my $machine = extract_param($param, 'machine'); + my $sync_time = extract_param($param, 'sync-time-on-resume'); my $get_root_param = sub { my $value = extract_param($param, $_[0]); @@ -3720,6 +3739,7 @@ __PACKAGE__->register_method({ timeout => $timeout, forcecpu => $force_cpu, 'nets-host-mtu' => $nets_host_mtu, + 'sync-time-on-resume' => $sync_time, }; PVE::QemuServer::vm_start($storecfg, $vmid, $params, $migrate_opts); @@ -4204,6 +4224,23 @@ __PACKAGE__->register_method({ description => "Do not check whether the VM is running, used" . " internally during migration. Only root may use this option.", }, + 'sync-time-on-resume' => { + type => 'boolean', + description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this resume only. Controls whether the guest's clock is" + . " synchronized after the RAM state is restored. The guest agent must be" + . " enabled.", + verbose_description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this resume only. When enabled, the 'guest-set-time' QEMU guest" + . " agent command is issued to synchronize the guest's clock to the host's current" + . " time after resuming the VM from a paused state. Without synchronization the" + . " guest's clock may still reflect the time from when the VM was paused. The" + . " time is only synchronized when the QEMU Guest Agent option is enabled in the" + . " guest's configuration and the agent is running inside of the guest.", + optional => 1, + }, }, }, returns => { @@ -4220,6 +4257,8 @@ __PACKAGE__->register_method({ my $vmid = extract_param($param, 'vmid'); + my $sync_time = extract_param($param, 'sync-time-on-resume'); + my $skiplock = extract_param($param, 'skiplock'); raise_param_exc({ skiplock => "Only root may use this option." }) if $skiplock && $authuser ne 'root@pam'; @@ -4250,10 +4289,14 @@ __PACKAGE__->register_method({ syslog('info', "resume VM $vmid: $upid\n"); if (!$to_disk_suspended) { - PVE::QemuServer::RunState::vm_resume($vmid, $skiplock, $nocheck); + PVE::QemuServer::RunState::vm_resume($vmid, $skiplock, $nocheck, $sync_time); } else { my $storecfg = PVE::Storage::config(); - PVE::QemuServer::vm_start($storecfg, $vmid, { skiplock => $skiplock }); + PVE::QemuServer::vm_start( + $storecfg, + $vmid, + { skiplock => $skiplock, 'sync-time-on-resume' => $sync_time }, + ); } return; @@ -6467,6 +6510,24 @@ __PACKAGE__->register_method({ optional => 1, default => 0, }, + 'sync-time-on-resume' => { + type => 'boolean', + description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this rollback only. Controls whether the guest's clock is" + . " synchronized after the RAM state is restored. The guest agent must be" + . " enabled.", + verbose_description => + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" + . " config for this rollback only. When enabled, the 'guest-set-time' QEMU guest" + . " agent command is issued to synchronize the guest's clock to the host's current" + . " time after rolling back to a snapshot that includes RAM. Without" + . " synchronization the guest's clock may still reflect the time from when the" + . " snapshot was taken. The time is only synchronized when the QEMU Guest Agent" + . " option is enabled in the guest's configuration and the agent is running" + . " inside of the guest.", + optional => 1, + }, }, }, returns => { @@ -6486,6 +6547,8 @@ __PACKAGE__->register_method({ my $snapname = extract_param($param, 'snapname'); + my $sync_time = extract_param($param, 'sync-time-on-resume'); + # vm_start is invoked directly from the worker, so its own permissions # predicate doesn't fire here - check VM.PowerMgmt up front so a user # with only VM.Snapshot.Rollback can't power the VM on as a side effect. @@ -6494,7 +6557,11 @@ __PACKAGE__->register_method({ my $realcmd = sub { PVE::Cluster::log_msg('info', $authuser, "rollback snapshot VM $vmid: $snapname"); - PVE::QemuConfig->snapshot_rollback($vmid, $snapname); + PVE::QemuConfig->snapshot_rollback( + $vmid, + $snapname, + { 'sync-time-on-resume' => $sync_time }, + ); if ($param->{start} && !PVE::QemuServer::Helpers::vm_running_locally($vmid)) { PVE::API2::Qemu->vm_start({ vmid => $vmid, node => $node }); @@ -7150,7 +7217,7 @@ __PACKAGE__->register_method({ }, 'resume' => sub { if (PVE::QemuServer::Helpers::vm_running_locally($state->{vmid})) { - PVE::QemuServer::RunState::vm_resume($state->{vmid}, 1, 1); + PVE::QemuServer::RunState::vm_resume($state->{vmid}, 1, 1, undef, 1); } else { die "VM $state->{vmid} not running\n"; } diff --git a/src/PVE/CLI/qm.pm b/src/PVE/CLI/qm.pm index f62fe036..ec3793ee 100755 --- a/src/PVE/CLI/qm.pm +++ b/src/PVE/CLI/qm.pm @@ -473,7 +473,7 @@ __PACKAGE__->register_method({ if (PVE::QemuServer::Helpers::vm_running_locally($vmid)) { # vm_resume with nocheck, since local node might not have processed config # move/rename yet - eval { PVE::QemuServer::RunState::vm_resume($vmid, 1, 1); }; + eval { PVE::QemuServer::RunState::vm_resume($vmid, 1, 1, undef, 1); }; if ($@) { $tunnel_write->("ERR: resume failed - $@"); } else { diff --git a/src/PVE/QemuConfig.pm b/src/PVE/QemuConfig.pm index c84f10f8..d69ab447 100644 --- a/src/PVE/QemuConfig.pm +++ b/src/PVE/QemuConfig.pm @@ -387,6 +387,11 @@ sub __snapshot_create_vol_snapshots_hook { next; } } + my $conf = $class->load_config($vmid); + if (PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent})) { + eval { PVE::QemuServer::Agent::guest_set_time($vmid); }; + warn "could not sync guest time after snapshot - $@" if $@; + } } } } @@ -470,8 +475,9 @@ sub __snapshot_rollback_hook { my ($class, $vmid, $conf, $snap, $prepare, $data) = @_; if ($prepare) { - # we save the machine of the current config + # we save the machine and agent of the current config $data->{oldmachine} = $conf->{machine}; + $data->{agent} = $conf->{agent}; } else { # if we have a 'runningmachine' entry in the snapshot we use that # for the forcemachine parameter, else we use the old logic @@ -535,7 +541,7 @@ sub __snapshot_rollback_vm_stop { } sub __snapshot_rollback_vm_start { - my ($class, $vmid, $vmstate, $data) = @_; + my ($class, $vmid, $vmstate, $data, $opts) = @_; my $storecfg = PVE::Storage::config(); my $params = { @@ -543,6 +549,8 @@ sub __snapshot_rollback_vm_start { forcemachine => $data->{forcemachine}, forcecpu => $data->{forcecpu}, 'nets-host-mtu' => $data->{'nets-host-mtu'}, + agent => $data->{agent}, + 'sync-time-on-resume' => $opts->{'sync-time-on-resume'}, }; PVE::QemuServer::vm_start($storecfg, $vmid, $params); } diff --git a/src/PVE/QemuMigrate.pm b/src/PVE/QemuMigrate.pm index 8176afe5..e24544a7 100644 --- a/src/PVE/QemuMigrate.pm +++ b/src/PVE/QemuMigrate.pm @@ -1778,6 +1778,9 @@ sub phase3_cleanup { $self->{errors} = 1; } } + $self->log('info', "guest clock sync requested, will run on target after resume") + if !$self->{errors} + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}); } if ( diff --git a/src/PVE/QemuServer.pm b/src/PVE/QemuServer.pm index 8149184e..55733ee1 100644 --- a/src/PVE/QemuServer.pm +++ b/src/PVE/QemuServer.pm @@ -5047,7 +5047,10 @@ sub vmconfig_update_agent { die "skip\n" if !$conf->{$opt}; - my $hotplug_options = { fstrim_cloned_disks => 1 }; + my $hotplug_options = { + fstrim_cloned_disks => 1, + 'sync-time-on-resume' => 1, + }; my $old_agent = parse_guest_agent($conf->{agent}); my $agent = parse_guest_agent($value); @@ -5910,6 +5913,19 @@ sub vm_start_nolock { ); } + # $resume = suspend-to-disk, $statefile = rollback with RAM. Rollback carries the snapshot's + # agent setting in params; hibernate resume falls back to the live config + if ($resume || ($statefile && !$migratedfrom)) { + my $agent = $params->{agent} // $conf->{agent}; + if (PVE::QemuServer::Agent::should_sync_time_on_resume( + $agent, + $params->{'sync-time-on-resume'}, + )) { + eval { PVE::QemuServer::Agent::guest_set_time($vmid); 1 }; + warn "could not sync guest time after resume from saved state - $@" if $@; + } + } + return $res; } diff --git a/src/PVE/QemuServer/Agent.pm b/src/PVE/QemuServer/Agent.pm index be6df443..5aaf76b2 100644 --- a/src/PVE/QemuServer/Agent.pm +++ b/src/PVE/QemuServer/Agent.pm @@ -4,6 +4,8 @@ use v5.36; use JSON; use MIME::Base64 qw(decode_base64 encode_base64); +use Time::HiRes qw(time); +use POSIX qw(strftime); use PVE::JSONSchema; @@ -52,6 +54,25 @@ our $agent_fmt = { optional => 1, default => 1, }, + # FIXME: MAJOR VERSION: 10.0.0 - reconsider setting option on as default + 'sync-time-on-resume' => { + description => + "Synchronize the guest's clock through QGA after operations that can leave" + . " the guest's clock behind.", + verbose_description => + "Whether to issue the 'guest-set-time' QEMU guest agent command to synchronize the" + . " guest's clock to the host's current time after operations that can leave the guest's" + . " clock behind. This happens when resuming from hibernation, resuming from a paused" + . " state, taking a snapshot, after a migration, after a backup in suspend mode, and" + . " after rolling back to a snapshot that includes RAM. In these cases the guest's" + . " clock may still reflect an earlier time. The time is only synchronized when the" + . " QEMU Guest Agent option is enabled in the guest's configuration and the agent is" + . " running inside of the guest. For resume, hibernation resume and snapshot rollback," + . " the synchronization can be overridden per operation.", + type => 'boolean', + optional => 1, + default => 0, + }, type => { description => "Select the agent type", type => 'string', @@ -332,4 +353,47 @@ sub guest_fs_freeze_applicable($agent_str, $vmid, $logfunc = undef) { return 1; } +=head3 should_sync_time_on_resume + +Returns whether the guest's clock should be synchronized to the host's via the QEMU Guest Agent +when the VM is resumed from saved state. Requires the guest agent to be enabled. An explicit +C<$override> (from an API parameter) takes precedence over the C<sync-time-on-resume> agent +config property when defined; otherwise the config value is used (defaulting to off). Does B<not> +check whether the agent is actually running. + +=cut + +sub should_sync_time_on_resume($agent_str, $override = undef) { + my $agent = parse_guest_agent($agent_str); + return 0 if !$agent->{enabled}; + return $override ? 1 : 0 if defined($override); + return $agent->{'sync-time-on-resume'} // 0; +} + +=head3 guest_set_time + +Sets the guest's clock to the current host time via the QEMU Guest Agent's +C<guest-set-time> command. The host time is passed explicitly (as nanoseconds +since the UNIX epoch) because the argument-less form reads the time from the +guest's RTC, which isn't supported on all platforms (e.g. Windows); passing +the value directly works regardless of guest OS. + +Returns 1 if the time was set, 0 if it was skipped because the agent is not running. +Dies if the agent returns an error. + +=cut + +sub guest_set_time($vmid, $quiet = undef) { + if (!qga_check_running($vmid, 1)) { + print "skipping guest set time - agent configured but not running?\n" if !$quiet; + return 0; + } + my $time_ns = time() * 1_000_000_000; + my $res = PVE::QemuServer::Monitor::mon_cmd($vmid, 'guest-set-time', time => int($time_ns)); + check_agent_error($res, "unable to set guest time"); + my $time_str = strftime("%F %T", localtime($time_ns / 1_000_000_000)); + print "synced guest clock via guest agent to $time_str\n" if !$quiet; + return 1; +} + 1; diff --git a/src/PVE/QemuServer/RunState.pm b/src/PVE/QemuServer/RunState.pm index 889ddd28..2588dabb 100644 --- a/src/PVE/QemuServer/RunState.pm +++ b/src/PVE/QemuServer/RunState.pm @@ -131,7 +131,7 @@ sub vm_suspend($vmid, $skiplock = undef, $includestate = undef, $statestorage = # $nocheck is set when called as part of a migration - in this context the # location of the config file (source or target node) is not deterministic, # since migration cannot wait for pmxcfs to process the rename -sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { +sub vm_resume($vmid, $skiplock = undef, $nocheck = undef, $sync_time = undef, $quiet = undef) { PVE::QemuConfig->lock_config( $vmid, @@ -183,6 +183,19 @@ sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { if $resume_cmd eq 'cont'; mon_cmd($vmid, $resume_cmd); + + # A VM in 'prelaunch' never ran, and a VM that shut down during a backup is reset + # above. Both boot fresh, so the clock is not behind and the agent is not up yet. + my $guest_was_running = !$reset && ($res->{status} // '') ne 'prelaunch'; + if ( + $resume_cmd eq 'cont' + && $guest_was_running + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}, + $sync_time) + ) { + eval { PVE::QemuServer::Agent::guest_set_time($vmid, $quiet); }; + warn "could not sync guest time after resume - $@" if $@; + } }, ); } diff --git a/src/PVE/VZDump/QemuServer.pm b/src/PVE/VZDump/QemuServer.pm index b1240f8f..292e7870 100644 --- a/src/PVE/VZDump/QemuServer.pm +++ b/src/PVE/VZDump/QemuServer.pm @@ -1185,20 +1185,41 @@ sub enforce_vm_running_for_backup { die $@ if $@; } +my sub sync_guest_time_after_resume { + my ($self, $vmid) = @_; + + my $agent_str = $self->{vmlist}->{$vmid}->{agent}; + return if !PVE::QemuServer::Agent::should_sync_time_on_resume($agent_str); + + my $synced = eval { PVE::QemuServer::Agent::guest_set_time($vmid, 1) }; + if (my $err = $@) { + $self->log('warn', "could not sync guest time - $err"); + } elsif ($synced) { + $self->loginfo("synced guest clock via guest agent"); + } else { + $self->loginfo("skipping guest time sync - agent configured but not running?"); + } +} + # resume VM again once in a clear state (stop mode backup of running VM) sub resume_vm_after_job_start { my ($self, $task, $vmid) = @_; return if !$self->{vm_was_running} || $self->{vm_was_paused}; - if (my $stoptime = $task->{vmstoptime}) { - my $delay = time() - $task->{vmstoptime}; + my $stoptime = $task->{vmstoptime}; + if ($stoptime) { + my $delay = time() - $stoptime; $task->{vmstoptime} = undef; # avoid printing 'online after ..' twice $self->loginfo("resuming VM again after $delay seconds"); } else { $self->loginfo("resuming VM again"); } mon_cmd($vmid, 'cont', timeout => 45); + + # only after a real vCPU pause: in stop mode the guest just booted, and $stoptime is + # cleared after the first call, so the error path does not sync twice + sync_guest_time_after_resume($self, $vmid) if $stoptime && $task->{mode} eq 'suspend'; } # stop again if VM was not running before -- 2.47.3 ^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind 2026-11-10 10:46 ` [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind Jakob Klocker @ 2026-11-13 14:10 ` Fiona Ebner 2026-11-16 12:21 ` Jakob Klocker 0 siblings, 1 reply; 10+ messages in thread From: Fiona Ebner @ 2026-11-13 14:10 UTC (permalink / raw) To: Jakob Klocker, pve-devel Needs a rebase Am 10.11.26 um 11:49 AM schrieb Jakob Klocker: > diff --git a/src/PVE/API2/Qemu.pm b/src/PVE/API2/Qemu.pm > index 7830d09b..df2076a6 100644 > --- a/src/PVE/API2/Qemu.pm > +++ b/src/PVE/API2/Qemu.pm > @@ -3599,6 +3599,24 @@ __PACKAGE__->register_method({ > description => 'Whether to migrate conntrack entries for running VMs.', > }, > 'nets-host-mtu' => get_standard_option('pve-qm-nets-host-mtu'), > + 'sync-time-on-resume' => { > + type => 'boolean', > + description => > + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" > + . " config for this start only. Controls whether the guest's clock is" Should explicitly state that this is for migration and hibernate and not regular start. > + . " synchronized after the RAM state is restored. The guest agent must be" > + . " enabled.", Nit: For this and all others: maybe we should drop the last sentence for the short description? It already is mentioned that it's about the guest agent setting. But fine either way. > + verbose_description => > + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" > + . " config for this start only. When enabled, the 'guest-set-time' QEMU guest" > + . " agent command is issued to synchronize the guest's clock to the host's current" > + . " time after the RAM state is restored, which happens here when starting a VM" > + . " that was hibernated (suspended to disk). Without synchronization the guest's" > + . " clock may still reflect the time from when the VM was hibernated. The time is" It's also for migration, or? > + . " only synchronized when the QEMU Guest Agent option is enabled in the guest's" > + . " configuration and the agent is running inside of the guest.", > + optional => 1, > + }, > }, > }, > returns => { ---snip 8<--- > diff --git a/src/PVE/QemuMigrate.pm b/src/PVE/QemuMigrate.pm > index 8176afe5..e24544a7 100644 > --- a/src/PVE/QemuMigrate.pm > +++ b/src/PVE/QemuMigrate.pm > @@ -1778,6 +1778,9 @@ sub phase3_cleanup { > $self->{errors} = 1; > } > } > + $self->log('info', "guest clock sync requested, will run on target after resume") > + if !$self->{errors} > + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}); Style nit: multiline post-if > } > > if ( > diff --git a/src/PVE/QemuServer.pm b/src/PVE/QemuServer.pm > index 8149184e..55733ee1 100644 > --- a/src/PVE/QemuServer.pm > +++ b/src/PVE/QemuServer.pm > @@ -5047,7 +5047,10 @@ sub vmconfig_update_agent { > > die "skip\n" if !$conf->{$opt}; > > - my $hotplug_options = { fstrim_cloned_disks => 1 }; > + my $hotplug_options = { > + fstrim_cloned_disks => 1, > + 'sync-time-on-resume' => 1, > + }; > > my $old_agent = parse_guest_agent($conf->{agent}); > my $agent = parse_guest_agent($value); > @@ -5910,6 +5913,19 @@ sub vm_start_nolock { > ); > } > > + # $resume = suspend-to-disk, $statefile = rollback with RAM. Rollback carries the snapshot's > + # agent setting in params; hibernate resume falls back to the live config > + if ($resume || ($statefile && !$migratedfrom)) { > + my $agent = $params->{agent} // $conf->{agent}; > + if (PVE::QemuServer::Agent::should_sync_time_on_resume( > + $agent, > + $params->{'sync-time-on-resume'}, > + )) { > + eval { PVE::QemuServer::Agent::guest_set_time($vmid); 1 }; Did you intend to pass 1 as an argument? > + warn "could not sync guest time after resume from saved state - $@" if $@; > + } > + } > + > return $res; > } > > diff --git a/src/PVE/QemuServer/Agent.pm b/src/PVE/QemuServer/Agent.pm > index be6df443..5aaf76b2 100644 > --- a/src/PVE/QemuServer/Agent.pm > +++ b/src/PVE/QemuServer/Agent.pm > @@ -4,6 +4,8 @@ use v5.36; > > use JSON; > use MIME::Base64 qw(decode_base64 encode_base64); > +use Time::HiRes qw(time); > +use POSIX qw(strftime); > > use PVE::JSONSchema; > > @@ -52,6 +54,25 @@ our $agent_fmt = { > optional => 1, > default => 1, > }, > + # FIXME: MAJOR VERSION: 10.0.0 - reconsider setting option on as default > + 'sync-time-on-resume' => { > + description => > + "Synchronize the guest's clock through QGA after operations that can leave" > + . " the guest's clock behind.", > + verbose_description => > + "Whether to issue the 'guest-set-time' QEMU guest agent command to synchronize the" > + . " guest's clock to the host's current time after operations that can leave the guest's" > + . " clock behind. This happens when resuming from hibernation, resuming from a paused" > + . " state, taking a snapshot, after a migration, after a backup in suspend mode, and" > + . " after rolling back to a snapshot that includes RAM. In these cases the guest's" > + . " clock may still reflect an earlier time. The time is only synchronized when the" > + . " QEMU Guest Agent option is enabled in the guest's configuration and the agent is" > + . " running inside of the guest. For resume, hibernation resume and snapshot rollback," > + . " the synchronization can be overridden per operation.", > + type => 'boolean', > + optional => 1, > + default => 0, > + }, > type => { > description => "Select the agent type", > type => 'string', > @@ -332,4 +353,47 @@ sub guest_fs_freeze_applicable($agent_str, $vmid, $logfunc = undef) { > return 1; > } > > +=head3 should_sync_time_on_resume > + > +Returns whether the guest's clock should be synchronized to the host's via the QEMU Guest Agent > +when the VM is resumed from saved state. Requires the guest agent to be enabled. An explicit This function can deal just fine with the situation where the agent is not enabled. Maybe state: "Returns 0 if the agent is not enabled.". > +C<$override> (from an API parameter) takes precedence over the C<sync-time-on-resume> agent > +config property when defined; otherwise the config value is used (defaulting to off). Does B<not> > +check whether the agent is actually running. > + > +=cut > + > +sub should_sync_time_on_resume($agent_str, $override = undef) { > + my $agent = parse_guest_agent($agent_str); > + return 0 if !$agent->{enabled}; > + return $override ? 1 : 0 if defined($override); > + return $agent->{'sync-time-on-resume'} // 0; > +} > + > +=head3 guest_set_time > + > +Sets the guest's clock to the current host time via the QEMU Guest Agent's > +C<guest-set-time> command. The host time is passed explicitly (as nanoseconds > +since the UNIX epoch) because the argument-less form reads the time from the > +guest's RTC, which isn't supported on all platforms (e.g. Windows); passing > +the value directly works regardless of guest OS. > + > +Returns 1 if the time was set, 0 if it was skipped because the agent is not running. > +Dies if the agent returns an error. Style nit: above used a length limit of 100 for wrapping, the first paragraph here 80, the second paragraph seems to be something in between. > + > +=cut > + > +sub guest_set_time($vmid, $quiet = undef) { > + if (!qga_check_running($vmid, 1)) { > + print "skipping guest set time - agent configured but not running?\n" if !$quiet; > + return 0; > + } > + my $time_ns = time() * 1_000_000_000; > + my $res = PVE::QemuServer::Monitor::mon_cmd($vmid, 'guest-set-time', time => int($time_ns)); > + check_agent_error($res, "unable to set guest time"); > + my $time_str = strftime("%F %T", localtime($time_ns / 1_000_000_000)); > + print "synced guest clock via guest agent to $time_str\n" if !$quiet; > + return 1; > +} > + > 1; > diff --git a/src/PVE/QemuServer/RunState.pm b/src/PVE/QemuServer/RunState.pm > index 889ddd28..2588dabb 100644 > --- a/src/PVE/QemuServer/RunState.pm > +++ b/src/PVE/QemuServer/RunState.pm > @@ -131,7 +131,7 @@ sub vm_suspend($vmid, $skiplock = undef, $includestate = undef, $statestorage = > # $nocheck is set when called as part of a migration - in this context the > # location of the config file (source or target node) is not deterministic, > # since migration cannot wait for pmxcfs to process the rename > -sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { > +sub vm_resume($vmid, $skiplock = undef, $nocheck = undef, $sync_time = undef, $quiet = undef) { > > PVE::QemuConfig->lock_config( > $vmid, > @@ -183,6 +183,19 @@ sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { > if $resume_cmd eq 'cont'; > > mon_cmd($vmid, $resume_cmd); > + > + # A VM in 'prelaunch' never ran, and a VM that shut down during a backup is reset > + # above. Both boot fresh, so the clock is not behind and the agent is not up yet. > + my $guest_was_running = !$reset && ($res->{status} // '') ne 'prelaunch'; Nit: could also skip 'shutdown' state for completeness. > + if ( > + $resume_cmd eq 'cont' > + && $guest_was_running > + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}, > + $sync_time) > + ) { > + eval { PVE::QemuServer::Agent::guest_set_time($vmid, $quiet); }; > + warn "could not sync guest time after resume - $@" if $@; > + } > }, > ); > } ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind 2026-11-13 14:10 ` Fiona Ebner @ 2026-11-16 12:21 ` Jakob Klocker 2026-11-17 8:57 ` Fiona Ebner 0 siblings, 1 reply; 10+ messages in thread From: Jakob Klocker @ 2026-11-16 12:21 UTC (permalink / raw) To: Fiona Ebner, pve-devel Thanks for the feedback! On Fri Nov 13, 2026 at 3:10 PM CET, Fiona Ebner wrote: > Needs a rebase ack > > Am 10.11.26 um 11:49 AM schrieb Jakob Klocker: >> diff --git a/src/PVE/API2/Qemu.pm b/src/PVE/API2/Qemu.pm >> index 7830d09b..df2076a6 100644 >> --- a/src/PVE/API2/Qemu.pm >> +++ b/src/PVE/API2/Qemu.pm >> @@ -3599,6 +3599,24 @@ __PACKAGE__->register_method({ >> description => 'Whether to migrate conntrack entries for running VMs.', >> }, >> 'nets-host-mtu' => get_standard_option('pve-qm-nets-host-mtu'), >> + 'sync-time-on-resume' => { >> + type => 'boolean', >> + description => >> + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" >> + . " config for this start only. Controls whether the guest's clock is" > > Should explicitly state that this is for migration and hibernate and not > regular start. Agreed about mentioning hibernate here; I'll make that explicit in v7. It is not used for migration though, see below. > >> + . " synchronized after the RAM state is restored. The guest agent must be" >> + . " enabled.", > > Nit: For this and all others: maybe we should drop the last sentence for > the short description? It already is mentioned that it's about the guest > agent setting. But fine either way. > ack, will remove >> + verbose_description => >> + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" >> + . " config for this start only. When enabled, the 'guest-set-time' QEMU guest" >> + . " agent command is issued to synchronize the guest's clock to the host's current" >> + . " time after the RAM state is restored, which happens here when starting a VM" >> + . " that was hibernated (suspended to disk). Without synchronization the guest's" >> + . " clock may still reflect the time from when the VM was hibernated. The time is" > > It's also for migration, or? > No. The migration target is started through this endpoint, but the RAM only arrives after vm_start returns AFAICT, so at that point the guest is not running yet and the agent can't respond. That's why the sync in vm_start_nolock is skipped when 'migratedfrom' is set. For migration, the clock is synced in vm_resume on the target, when the source sends the 'resume' command through the migration tunnel after the switchover. That always uses the value from the agent config. >> + . " only synchronized when the QEMU Guest Agent option is enabled in the guest's" >> + . " configuration and the agent is running inside of the guest.", >> + optional => 1, >> + }, >> }, >> }, >> returns => { > > ---snip 8<--- > >> diff --git a/src/PVE/QemuMigrate.pm b/src/PVE/QemuMigrate.pm >> index 8176afe5..e24544a7 100644 >> --- a/src/PVE/QemuMigrate.pm >> +++ b/src/PVE/QemuMigrate.pm >> @@ -1778,6 +1778,9 @@ sub phase3_cleanup { >> $self->{errors} = 1; >> } >> } >> + $self->log('info', "guest clock sync requested, will run on target after resume") >> + if !$self->{errors} >> + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}); > > Style nit: multiline post-if ack > >> } >> >> if ( >> diff --git a/src/PVE/QemuServer.pm b/src/PVE/QemuServer.pm >> index 8149184e..55733ee1 100644 >> --- a/src/PVE/QemuServer.pm >> +++ b/src/PVE/QemuServer.pm >> @@ -5047,7 +5047,10 @@ sub vmconfig_update_agent { >> >> die "skip\n" if !$conf->{$opt}; >> >> - my $hotplug_options = { fstrim_cloned_disks => 1 }; >> + my $hotplug_options = { >> + fstrim_cloned_disks => 1, >> + 'sync-time-on-resume' => 1, >> + }; >> >> my $old_agent = parse_guest_agent($conf->{agent}); >> my $agent = parse_guest_agent($value); >> @@ -5910,6 +5913,19 @@ sub vm_start_nolock { >> ); >> } >> >> + # $resume = suspend-to-disk, $statefile = rollback with RAM. Rollback carries the snapshot's >> + # agent setting in params; hibernate resume falls back to the live config >> + if ($resume || ($statefile && !$migratedfrom)) { >> + my $agent = $params->{agent} // $conf->{agent}; >> + if (PVE::QemuServer::Agent::should_sync_time_on_resume( >> + $agent, >> + $params->{'sync-time-on-resume'}, >> + )) { >> + eval { PVE::QemuServer::Agent::guest_set_time($vmid); 1 }; > > Did you intend to pass 1 as an argument? > No, this shouldn't be `quiet`. I'm not sure how this sneaked in. Will be removed in v7. > >> + warn "could not sync guest time after resume from saved state - $@" if $@; >> + } >> + } >> + >> return $res; >> } >> >> diff --git a/src/PVE/QemuServer/Agent.pm b/src/PVE/QemuServer/Agent.pm >> index be6df443..5aaf76b2 100644 >> --- a/src/PVE/QemuServer/Agent.pm >> +++ b/src/PVE/QemuServer/Agent.pm >> @@ -4,6 +4,8 @@ use v5.36; >> >> use JSON; >> use MIME::Base64 qw(decode_base64 encode_base64); >> +use Time::HiRes qw(time); >> +use POSIX qw(strftime); >> >> use PVE::JSONSchema; >> >> @@ -52,6 +54,25 @@ our $agent_fmt = { >> optional => 1, >> default => 1, >> }, >> + # FIXME: MAJOR VERSION: 10.0.0 - reconsider setting option on as default >> + 'sync-time-on-resume' => { >> + description => >> + "Synchronize the guest's clock through QGA after operations that can leave" >> + . " the guest's clock behind.", >> + verbose_description => >> + "Whether to issue the 'guest-set-time' QEMU guest agent command to synchronize the" >> + . " guest's clock to the host's current time after operations that can leave the guest's" >> + . " clock behind. This happens when resuming from hibernation, resuming from a paused" >> + . " state, taking a snapshot, after a migration, after a backup in suspend mode, and" >> + . " after rolling back to a snapshot that includes RAM. In these cases the guest's" >> + . " clock may still reflect an earlier time. The time is only synchronized when the" >> + . " QEMU Guest Agent option is enabled in the guest's configuration and the agent is" >> + . " running inside of the guest. For resume, hibernation resume and snapshot rollback," >> + . " the synchronization can be overridden per operation.", >> + type => 'boolean', >> + optional => 1, >> + default => 0, >> + }, >> type => { >> description => "Select the agent type", >> type => 'string', >> @@ -332,4 +353,47 @@ sub guest_fs_freeze_applicable($agent_str, $vmid, $logfunc = undef) { >> return 1; >> } >> >> +=head3 should_sync_time_on_resume >> + >> +Returns whether the guest's clock should be synchronized to the host's via the QEMU Guest Agent >> +when the VM is resumed from saved state. Requires the guest agent to be enabled. An explicit > > This function can deal just fine with the situation where the agent is > not enabled. Maybe state: "Returns 0 if the agent is not enabled.". > Agreed, will change in v7 >> +C<$override> (from an API parameter) takes precedence over the C<sync-time-on-resume> agent >> +config property when defined; otherwise the config value is used (defaulting to off). Does B<not> >> +check whether the agent is actually running. >> + >> +=cut >> + >> +sub should_sync_time_on_resume($agent_str, $override = undef) { >> + my $agent = parse_guest_agent($agent_str); >> + return 0 if !$agent->{enabled}; >> + return $override ? 1 : 0 if defined($override); >> + return $agent->{'sync-time-on-resume'} // 0; >> +} >> + >> +=head3 guest_set_time >> + >> +Sets the guest's clock to the current host time via the QEMU Guest Agent's >> +C<guest-set-time> command. The host time is passed explicitly (as nanoseconds >> +since the UNIX epoch) because the argument-less form reads the time from the >> +guest's RTC, which isn't supported on all platforms (e.g. Windows); passing >> +the value directly works regardless of guest OS. >> + >> +Returns 1 if the time was set, 0 if it was skipped because the agent is not running. >> +Dies if the agent returns an error. > > Style nit: above used a length limit of 100 for wrapping, the first > paragraph here 80, the second paragraph seems to be something in between. > ack >> + >> +=cut >> + >> +sub guest_set_time($vmid, $quiet = undef) { >> + if (!qga_check_running($vmid, 1)) { >> + print "skipping guest set time - agent configured but not running?\n" if !$quiet; >> + return 0; >> + } >> + my $time_ns = time() * 1_000_000_000; >> + my $res = PVE::QemuServer::Monitor::mon_cmd($vmid, 'guest-set-time', time => int($time_ns)); >> + check_agent_error($res, "unable to set guest time"); >> + my $time_str = strftime("%F %T", localtime($time_ns / 1_000_000_000)); >> + print "synced guest clock via guest agent to $time_str\n" if !$quiet; >> + return 1; >> +} >> + >> 1; >> diff --git a/src/PVE/QemuServer/RunState.pm b/src/PVE/QemuServer/RunState.pm >> index 889ddd28..2588dabb 100644 >> --- a/src/PVE/QemuServer/RunState.pm >> +++ b/src/PVE/QemuServer/RunState.pm >> @@ -131,7 +131,7 @@ sub vm_suspend($vmid, $skiplock = undef, $includestate = undef, $statestorage = >> # $nocheck is set when called as part of a migration - in this context the >> # location of the config file (source or target node) is not deterministic, >> # since migration cannot wait for pmxcfs to process the rename >> -sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { >> +sub vm_resume($vmid, $skiplock = undef, $nocheck = undef, $sync_time = undef, $quiet = undef) { >> >> PVE::QemuConfig->lock_config( >> $vmid, >> @@ -183,6 +183,19 @@ sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { >> if $resume_cmd eq 'cont'; >> >> mon_cmd($vmid, $resume_cmd); >> + >> + # A VM in 'prelaunch' never ran, and a VM that shut down during a backup is reset >> + # above. Both boot fresh, so the clock is not behind and the agent is not up yet. >> + my $guest_was_running = !$reset && ($res->{status} // '') ne 'prelaunch'; > > Nit: could also skip 'shutdown' state for completeness. > That's already covered: $reset is only set for the `shutdown` state, and $guest_was_running checks !$reset. I'll check the status directly in v7, so this is more obvious. Something along the lines: my $status = $res->{status} // ''; my $guest_was_running = $status ne 'prelaunch' && $status ne 'shutdown'; >> + if ( >> + $resume_cmd eq 'cont' >> + && $guest_was_running >> + && PVE::QemuServer::Agent::should_sync_time_on_resume($conf->{agent}, >> + $sync_time) >> + ) { >> + eval { PVE::QemuServer::Agent::guest_set_time($vmid, $quiet); }; >> + warn "could not sync guest time after resume - $@" if $@; >> + } >> }, >> ); >> } ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind 2026-11-16 12:21 ` Jakob Klocker @ 2026-11-17 8:57 ` Fiona Ebner 0 siblings, 0 replies; 10+ messages in thread From: Fiona Ebner @ 2026-11-17 8:57 UTC (permalink / raw) To: Jakob Klocker, pve-devel Am 16.11.26 um 1:21 PM schrieb Jakob Klocker: > Thanks for the feedback! > On Fri Nov 13, 2026 at 3:10 PM CET, Fiona Ebner wrote: >> Am 10.11.26 um 11:49 AM schrieb Jakob Klocker: >>> + verbose_description => >>> + "Override the 'sync-time-on-resume' setting from the VM's guest agent (QGA)" >>> + . " config for this start only. When enabled, the 'guest-set-time' QEMU guest" >>> + . " agent command is issued to synchronize the guest's clock to the host's current" >>> + . " time after the RAM state is restored, which happens here when starting a VM" >>> + . " that was hibernated (suspended to disk). Without synchronization the guest's" >>> + . " clock may still reflect the time from when the VM was hibernated. The time is" >> >> It's also for migration, or? >> > > No. The migration target is started through this endpoint, but the RAM > only arrives after vm_start returns AFAICT, so at that point the guest > is not running yet and the agent can't respond. That's why the sync in > vm_start_nolock is skipped when 'migratedfrom' is set. > > For migration, the clock is synced in vm_resume on the target, when > the source sends the 'resume' command through the migration tunnel > after the switchover. That always uses the value from the agent config. > Right, thanks for the clarification! >>> @@ -183,6 +183,19 @@ sub vm_resume($vmid, $skiplock = undef, $nocheck = undef) { >>> if $resume_cmd eq 'cont'; >>> >>> mon_cmd($vmid, $resume_cmd); >>> + >>> + # A VM in 'prelaunch' never ran, and a VM that shut down during a backup is reset >>> + # above. Both boot fresh, so the clock is not behind and the agent is not up yet. >>> + my $guest_was_running = !$reset && ($res->{status} // '') ne 'prelaunch'; >> >> Nit: could also skip 'shutdown' state for completeness. >> > > That's already covered: $reset is only set for the `shutdown` state, > and $guest_was_running checks !$reset. I'll check the status directly > in v7, so this is more obvious. Something along the lines: > > my $status = $res->{status} // ''; > my $guest_was_running = $status ne 'prelaunch' && $status ne 'shutdown'; Thanks! ^ permalink raw reply [flat|nested] 10+ messages in thread
* superseded: [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA 2026-11-10 10:45 [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker ` (2 preceding siblings ...) 2026-11-10 10:46 ` [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind Jakob Klocker @ 2026-11-17 12:41 ` Jakob Klocker 3 siblings, 0 replies; 10+ messages in thread From: Jakob Klocker @ 2026-11-17 12:41 UTC (permalink / raw) To: Jakob Klocker, pve-devel Superseded-by: https://lore.proxmox.com/all/20261117123932.995999-1-j.klocker@proxmox.com/T/#u ^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2026-11-17 12:41 UTC | newest] Thread overview: 10+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2026-11-10 10:45 [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker 2026-11-10 10:45 ` [PATCH pve-guest-common v6 1/3] AbstractConfig: allow passing options to snapshot_rollback Jakob Klocker 2026-11-13 14:10 ` Fiona Ebner 2026-11-16 12:23 ` Jakob Klocker 2026-11-10 10:46 ` [PATCH pve-manager v6 2/3] fix #5032: ui: qemu agent: add sync-time-on-resume option Jakob Klocker 2026-11-10 10:46 ` [PATCH qemu-server v6 3/3] fix #5032: agent: sync guest time via QGA when the clock falls behind Jakob Klocker 2026-11-13 14:10 ` Fiona Ebner 2026-11-16 12:21 ` Jakob Klocker 2026-11-17 8:57 ` Fiona Ebner 2026-11-17 12:41 ` superseded: [PATCH guest-common/manager/qemu-server v6 0/3] fix #5032: add guest time sync via QGA Jakob Klocker
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox