[PATCH storage v3 0/2] fix #7598: qemu-img: increase timeout for image resize

[PATCH storage v3 0/2] fix #7598: qemu-img: increase timeout for image resize

[Jakob Klocker]

Image resizing currently has a timeout of 10 seconds, which can fail
on slow disks and leave the VM in a state where the disk is updated,
but the VM config isn't. 

Since there is no deterministic way to check, with less overhead, if
this is the case, increasing the timeout is the most pragmatic
solution. The resize is only called in a worker context anyway,
therefore this patch series mirrors the ZFS timeout: it increases
the timeout in a worker context to 1 hour; as a fallback for a 
non-worker context the timeout stays 10 seconds.

Additionally, a minor refactoring: the timeout argument was dropped at
the two call sites of qemu_img_resize, since both passed the default
value of 10 seconds.

Changes since v2
----------------
- restore timeout parameter since the function can be used by
third-party plugins (Thanks @Fiona & @Elias)

Changes since v1
-----------------
- drop the timeout parameter in qemu_img_resize
- increase the timeout instead of querying the disk information
(see v1 for more info. Thanks @Fabian)

v1: https://lore.proxmox.com/pve-devel/20260603082557.25359-1-j.klocker@proxmox.com/t/#u

Link: https://bugzilla.proxmox.com/show_bug.cgi?id=7598


pve-storage:

Jakob Klocker (2):
  qemu-img: drop redundant timeout argument at resize call
  fix #7598: qemu-img: increase timeout for image resizing in worker
    context

 src/PVE/Storage/Common.pm    | 6 ++++--
 src/PVE/Storage/LVMPlugin.pm | 2 +-
 src/PVE/Storage/Plugin.pm    | 2 +-
 3 files changed, 6 insertions(+), 4 deletions(-)


Summary over all repositories:
  3 files changed, 6 insertions(+), 4 deletions(-)

-- 
Generated by murpp 0.12.0

[PATCH pve-storage v3 1/2] qemu-img: drop redundant timeout argument at resize call

[Jakob Klocker]

`qemu_img_resize` already defaults to a timeout of 10 seconds when none
is passed, so passing 10 explicitly is redundant.

Signed-off-by: Jakob Klocker <j.klocker@proxmox.com>
---
 src/PVE/Storage/LVMPlugin.pm | 2 +-
 src/PVE/Storage/Plugin.pm    | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/PVE/Storage/LVMPlugin.pm b/src/PVE/Storage/LVMPlugin.pm
index b8646c2..768ff95 100644
--- a/src/PVE/Storage/LVMPlugin.pm
+++ b/src/PVE/Storage/LVMPlugin.pm
@@ -1081,7 +1081,7 @@ sub volume_resize {
 
     if (!$running && $format eq 'qcow2') {
         my $preallocation = PVE::Storage::Plugin::preallocation_cmd_opt($scfg, $format);
-        PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation, 10);
+        PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation);
     }
 
     return 1;
diff --git a/src/PVE/Storage/Plugin.pm b/src/PVE/Storage/Plugin.pm
index 4f69f9b..082355d 100644
--- a/src/PVE/Storage/Plugin.pm
+++ b/src/PVE/Storage/Plugin.pm
@@ -1373,7 +1373,7 @@ sub volume_resize {
     my $format = ($class->parse_volname($volname))[6];
 
     my $preallocation = preallocation_cmd_opt($scfg, $format);
-    PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation, 10);
+    PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation);
 
     return undef;
 }
-- 
2.47.3

Re: [PATCH pve-storage v3 1/2] qemu-img: drop redundant timeout argument at resize call

[Elias Huhsovitz]

Last patch (v2) I also didn't consider that thrid-party plugins would
call qemu_img_resize. Therefore keeping the $timeout parameter is
definitely the correct choice.

Reviewed-by: Elias Huhsovitz <e.huhsovitz@proxmox.com>
Tested-by: Elias Huhsovitz <e.huhsovitz@proxmox.com>

On Tue Aug 11, 2026 at 9:37 AM CEST, Jakob Klocker wrote:
> `qemu_img_resize` already defaults to a timeout of 10 seconds when none
> is passed, so passing 10 explicitly is redundant.
>
> Signed-off-by: Jakob Klocker <j.klocker@proxmox.com>
> ---
>  src/PVE/Storage/LVMPlugin.pm | 2 +-
>  src/PVE/Storage/Plugin.pm    | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/src/PVE/Storage/LVMPlugin.pm b/src/PVE/Storage/LVMPlugin.pm
> index b8646c2..768ff95 100644
> --- a/src/PVE/Storage/LVMPlugin.pm
> +++ b/src/PVE/Storage/LVMPlugin.pm
> @@ -1081,7 +1081,7 @@ sub volume_resize {
>  
>      if (!$running && $format eq 'qcow2') {
>          my $preallocation = PVE::Storage::Plugin::preallocation_cmd_opt($scfg, $format);
> -        PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation, 10);
> +        PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation);
>      }
>  
>      return 1;
> diff --git a/src/PVE/Storage/Plugin.pm b/src/PVE/Storage/Plugin.pm
> index 4f69f9b..082355d 100644
> --- a/src/PVE/Storage/Plugin.pm
> +++ b/src/PVE/Storage/Plugin.pm
> @@ -1373,7 +1373,7 @@ sub volume_resize {
>      my $format = ($class->parse_volname($volname))[6];
>  
>      my $preallocation = preallocation_cmd_opt($scfg, $format);
> -    PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation, 10);
> +    PVE::Storage::Common::qemu_img_resize($path, $format, $size, $preallocation);
>  
>      return undef;
>  }

[PATCH pve-storage v3 2/2] fix #7598: qemu-img: increase timeout for image resizing in worker context

[Jakob Klocker]

Mirror ZFS behaviour and increase timeout to an hour when in a worker
context. This fixes issues some users [0] have with slow disks where a
resize would fail.

[0] https://bugzilla.proxmox.com/show_bug.cgi?id=7598

Signed-off-by: Jakob Klocker <j.klocker@proxmox.com>
---
 src/PVE/Storage/Common.pm | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/src/PVE/Storage/Common.pm b/src/PVE/Storage/Common.pm
index 3932aee..1c3fdad 100644
--- a/src/PVE/Storage/Common.pm
+++ b/src/PVE/Storage/Common.pm
@@ -3,6 +3,7 @@ package PVE::Storage::Common;
 use v5.36;
 
 use PVE::JSONSchema;
+use PVE::RPCEnvironment;
 use PVE::Syscall;
 use PVE::Tools qw(run_command);
 
@@ -262,8 +263,8 @@ sub qemu_img_measure {
     qemu_img_resize($path, $format, $size, $preallocation, $timeout)
 
 Resize a qemu image C<$path> with format C<$format> to a target Kb size C<$size>.
-Default timeout C<$timeout> is 10s if not specified.
 C<$preallocation> allows to specify the preallocation option for the resize operation.
+If no C<$timeout> is provided, it defaults to 10 seconds, or 1 hour when running in a worker context.
 
 =cut
 
@@ -276,7 +277,8 @@ sub qemu_img_resize {
     push $cmd->@*, "--preallocation=$preallocation" if $preallocation;
     push $cmd->@*, '-f', $format, $path, $size;
 
-    $timeout = 10 if !$timeout;
+    $timeout = PVE::RPCEnvironment->is_worker() ? 60 * 60 : 10
+        if !$timeout;
     run_command($cmd, timeout => $timeout);
 }
 
-- 
2.47.3

Re: [PATCH pve-storage v3 2/2] fix #7598: qemu-img: increase timeout for image resizing in worker context

[Elias Huhsovitz]

Tested again using 

pvesh set /nodes/{node}/qemu/{vmid}/resize --disk scsi0 --size +{num}G

and the GUI option VM->Disk Action->Resize

Which correctly set the $timeout=3600 in the worker context as intended.

Reviewed-by: Elias Huhsovitz <e.huhsovitz@proxmox.com>
Tested-by: Elias Huhsovitz <e.huhsovitz@proxmox.com>

On Tue Aug 11, 2026 at 9:37 AM CEST, Jakob Klocker wrote:
> Mirror ZFS behaviour and increase timeout to an hour when in a worker
> context. This fixes issues some users [0] have with slow disks where a
> resize would fail.
>
> [0] https://bugzilla.proxmox.com/show_bug.cgi?id=7598
>
> Signed-off-by: Jakob Klocker <j.klocker@proxmox.com>
> ---
>  src/PVE/Storage/Common.pm | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/src/PVE/Storage/Common.pm b/src/PVE/Storage/Common.pm
> index 3932aee..1c3fdad 100644
> --- a/src/PVE/Storage/Common.pm
> +++ b/src/PVE/Storage/Common.pm
> @@ -3,6 +3,7 @@ package PVE::Storage::Common;
>  use v5.36;
>  
>  use PVE::JSONSchema;
> +use PVE::RPCEnvironment;
>  use PVE::Syscall;
>  use PVE::Tools qw(run_command);
>  
> @@ -262,8 +263,8 @@ sub qemu_img_measure {
>      qemu_img_resize($path, $format, $size, $preallocation, $timeout)
>  
>  Resize a qemu image C<$path> with format C<$format> to a target Kb size C<$size>.
> -Default timeout C<$timeout> is 10s if not specified.
>  C<$preallocation> allows to specify the preallocation option for the resize operation.
> +If no C<$timeout> is provided, it defaults to 10 seconds, or 1 hour when running in a worker context.
>  
>  =cut
>  
> @@ -276,7 +277,8 @@ sub qemu_img_resize {
>      push $cmd->@*, "--preallocation=$preallocation" if $preallocation;
>      push $cmd->@*, '-f', $format, $path, $size;
>  
> -    $timeout = 10 if !$timeout;
> +    $timeout = PVE::RPCEnvironment->is_worker() ? 60 * 60 : 10
> +        if !$timeout;
>      run_command($cmd, timeout => $timeout);
>  }
>  

applied: [PATCH storage v3 0/2] fix #7598: qemu-img: increase timeout for image resize

[Fiona Ebner]

On Tue, 11 Aug 2026 09:37:38 +0200, Jakob Klocker wrote:
> Image resizing currently has a timeout of 10 seconds, which can fail
> on slow disks and leave the VM in a state where the disk is updated,
> but the VM config isn't.
> 
> Since there is no deterministic way to check, with less overhead, if
> this is the case, increasing the timeout is the most pragmatic
> solution. The resize is only called in a worker context anyway,
> therefore this patch series mirrors the ZFS timeout: it increases
> the timeout in a worker context to 1 hour; as a fallback for a
> non-worker context the timeout stays 10 seconds.
> 
> [...]

Applied, thanks! In the second patch, I made two small changes:
* style: avoid combining ternary conditional operator and post-if
* wrap doc comment to avoid line longer than 100 characters

[1/2] qemu-img: drop redundant timeout argument at resize call
      commit: 1b930a64a67ba21b7e3f4d3cfb578f240c71b661
[2/2] fix #7598: qemu-img: increase timeout for image resizing in worker context
      commit: e27a8036f5dc9d2d09e427b055264d97725566d7