From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from gate001.proxmox.com (gate001.proxmox.com [IPv6:2a0f:8001:1:32::40]) by lore.proxmox.com (Postfix) with ESMTPS id DF6871FF0DE for ; Thu, 19 Nov 2026 12:45:57 +0100 (CET) Received: from gate001.proxmox.com (localhost.localdomain [127.0.0.1]) by gate001.proxmox.com (Proxmox) with ESMTP id A863D216EB; Thu, 19 Nov 2026 12:45:55 +0100 (CET) From: Maximiliano Sandoval To: Jakob Klocker Subject: Re: [PATCH pve-docs] qm: guest agent: document sync-time-on-resume option In-Reply-To: <20261119103924.156068-1-j.klocker@proxmox.com> (Jakob Klocker's message of "Thu, 19 Nov 2026 11:39:24 +0100") References: <20261119103924.156068-1-j.klocker@proxmox.com> User-Agent: mu4e 1.12.9; emacs 30.1 Date: Thu, 19 Nov 2026 12:45:50 +0100 Message-ID: MIME-Version: 1.0 Content-Type: text/plain X-Bm-Milter-Handled: 55990f41-d878-4baa-be0a-ee34c49e34d2 X-Bm-Transport-Timestamp: 1795088750623 X-SPAM-LEVEL: Spam detection results: 0 AWL 0.711 Adjusted score from AWL reputation of From: address DMARC_MISSING 0.1 Missing DMARC policy KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment (newer systems) RCVD_IN_DNSWL_MED -2.3 Sender listed at https://www.dnswl.org/, medium trust SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record SPF_PASS -0.001 SPF: sender matches SPF record Message-ID-Hash: CGWXTPSE7ZJDNXD253SFSV63NFYCEO72 X-Message-ID-Hash: CGWXTPSE7ZJDNXD253SFSV63NFYCEO72 X-MailFrom: m.sandoval@proxmox.com X-Mailman-Rule-Misses: dmarc-mitigation; no-senders; approved; loop; banned-address; emergency; member-moderation; nonmember-moderation; administrivia; implicit-dest; max-recipients; max-size; news-moderation; no-subject; digests; suspicious-header CC: pve-devel@lists.proxmox.com X-Mailman-Version: 3.3.10 Precedence: list List-Id: Proxmox VE development discussion List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: Jakob Klocker writes: > Add a section about the new 'sync-time-on-resume' agent option to the > QEMU Guest Agent docs. > > Signed-off-by: Jakob Klocker > --- > qm.adoc | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++ > 1 file changed, 62 insertions(+) > > diff --git a/qm.adoc b/qm.adoc > index 18a4fbe..6e86b60 100644 > --- a/qm.adoc > +++ b/qm.adoc > @@ -1660,6 +1660,68 @@ IMPORTANT: Disabling this option can potentially lead to backups with > inconsistent filesystems. Therefore, adapting the QEMU Guest Agent > configuration in the guest is the preferred option. > > +[[qm_qga_time_sync]] > +Guest Clock Synchronization > +^^^^^^^^^^^^^^^^^^^^^^^^^^^ > + > +While a VM is paused, the guest's clock does not advance. The same applies when > +a VM continues from a previously saved state, for example after hibernation or > +after rolling back to a snapshot that includes RAM. Once the VM runs again, the Perhaps it is worth mentioning that this can also happen in certain scenarios if the last delta of migration takes a long time to be sent. We had a couple of support cases where that happened. cc @laurentiu > +guest's clock lags behind the host's time. A time synchronization service in > +the guest, such as NTP, may correct a large offset only slowly or not at all, > +depending on its configuration. Until then, time-sensitive services can > +misbehave. > + > +With the `sync-time-on-resume` option enabled, {pve} sets the guest's clock to nit: Perhaps: With the `sync-time-on-resume` QEMU Guest Agent option enabled, {pve} sets the guest's clock to it is not clear where to enable it otherwise (this can allow making the paragraph below a bit leaner). > +the host's current time via the `guest-set-time` QEMU Guest Agent command right nit: Perhaps add a link to https://www.qemu.org/docs/master/interop/qemu-ga-ref.html#command-QGA-qapi-schema.guest-get-time. > +after the following operations: > + > +* resuming a paused VM > +* resuming a hibernated VM > +* rolling back to a snapshot that includes RAM > +* taking a snapshot of a running VM > +* live migrating a VM to another node > +* performing a backup in suspend mode > + > +The option is disabled by default. To enable it in the GUI, open the 'Options' > +panel of the virtual machine, then the 'QEMU Guest Agent' dialog, and enable > +the 'Synchronize guest clock after resume, snapshot, rollback or migration' > +option. On the command line, add `sync-time-on-resume=1` to the `agent` option, > +keeping any other agent options you have already set, for example: > + > +---- > +# qm set --agent 1,sync-time-on-resume=1 > +---- nit: I do not have anything against this, but to me it seems a bit over-verbose. The documentation rarely explains in such detail how to set a setting. > + > +.Overriding the Option for a Single Operation > +For a single operation, the configured value can be overridden with the > +`--sync-time-on-resume` parameter. This is supported when resuming a paused or > +hibernated VM (`qm resume`), starting a hibernated VM (`qm start`), and rolling > +back to a snapshot with RAM (`qm rollback`). With `1`, the clock is > +synchronized even if the option is disabled; with `0`, it is not > +synchronized even if the option is enabled. For example, to resume a > +VM without synchronizing its clock: > + > +---- > +# qm resume --sync-time-on-resume 0 > +---- > + > +Taking a snapshot, live migrations and backups always use the configured value. > + > +.Rolling Back to a Snapshot > +When rolling back to a snapshot with RAM, {pve} decides whether to synchronize > +the clock as follows: > + > +* If the `--sync-time-on-resume` parameter is given, its value is used. > +* Otherwise, if the guest agent is enabled in the configuration before the > + rollback, its `sync-time-on-resume` setting is used (disabled if not set). > + This way, the current setting also applies to snapshots taken before the > + option was enabled. > +* Otherwise, the `sync-time-on-resume` setting stored in the snapshot is used. > + > +In any case, the guest agent must be enabled in the snapshot's configuration, nit: And the agent must be running inside of the guest. > +since the restored VM runs with the hardware configuration of the snapshot. > + > Troubleshooting > ^^^^^^^^^^^^^^^ -- Maximiliano