[pve-devel] [PATCH qemu-server v2] vm start: set higher timeout if using PCI passthrough

[Friedrich Weber]

The default VM startup timeout is `max(30, VM memory in GiB)` seconds.
Multiple reports in the forum [0] [1] and the bug tracker [2] suggest
this is too short when using PCI passthrough with a large amount of VM
memory, since QEMU needs to map the whole memory during startup (see
comment #2 in [2]). As a result, VM startup fails with "got timeout".

To work around this, set a larger default timeout if at least one PCI
device is passed through. The question remains how to choose an
appropriate timeout. Users reported the following startup times:

ref | RAM | time  | ratio (s/GiB)
---------------------------------
[1] | 60G |  135s |  2.25
[1] | 70G |  157s |  2.24
[1] | 80G |  277s |  3.46
[2] | 65G |  213s |  3.28
[2] | 96G | >290s | >3.02

The data does not really indicate any simple (e.g. linear)
relationship between RAM and startup time (even data from the same
source). However, to keep the heuristic simple, assume linear growth
and multiply the default timeout by 4 if at least one `hostpci[n]`
option is present, obtaining `4 * max(30, VM memory in GiB)`. This
covers all cases above, and should still leave some headroom.

[0]: https://forum.proxmox.com/threads/83765/post-552071
[1]: https://forum.proxmox.com/threads/126398/post-592826
[2]: https://bugzilla.proxmox.com/show_bug.cgi?id=3502

Suggested-by: Fiona Ebner <f.ebner at proxmox.com>
Signed-off-by: Friedrich Weber <f.weber at proxmox.com>
---

Notes:
    changes since v1 (was called "vm start: set minimum timeout of 300s if
    using PCI passthrough", 20230503133723.165739-1-f.weber at proxmox.com):
    * Use a constant multiplier as suggested by Fiona (thx!)
    
    Another workaround is offered by an unapplied patch series [3] of bug
    3502 [2] that makes it possible to set VM-specific timeouts (also in
    the GUI). Users could use this option to manually set a higher
    timeout for VMs that use PCI passthrough. However, it is not
    immediately obvious that a higher timeout is necessary when using
    PCI passthrough. Since the problem seems to come up somewhat
    frequently, I think it makes sense to have the heuristic choose a
    higher timeout by default.
    
    As discussed in v1, I'll also pick up the patch series to allow users
    to set custom timeouts [3], also to offer a workaround for cases where
    the new heuristic chooses a timeout that is still too short.
    
    [2]: https://bugzilla.proxmox.com/show_bug.cgi?id=3502
    [3]: https://lists.proxmox.com/pipermail/pve-devel/2023-January/055352.html

 PVE/QemuServer/Helpers.pm | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/PVE/QemuServer/Helpers.pm b/PVE/QemuServer/Helpers.pm
index 8817427..0afb631 100644
--- a/PVE/QemuServer/Helpers.pm
+++ b/PVE/QemuServer/Helpers.pm
@@ -152,6 +152,13 @@ sub config_aware_timeout {
 	$timeout = int($memory/1024);
     }
 
+    # When using PCI passthrough, users reported much higher startup times,
+    # growing with the amount of memory configured. Constant factor chosen
+    # based on user reports.
+    if (grep(/^hostpci[0-9]+$/, keys %$config)) {
+	$timeout *= 4;
+    }
+
     if ($is_suspended && $timeout < 300) {
 	$timeout = 300;
     }
-- 
2.39.2