[pve-devel] [PATCH qemu-server] fix #3329: turn on cache=writeback for efidisks on rbd

[pve-devel] [PATCH qemu-server] fix #3329: turn on cache=writeback for efidisks on rbd

[Dominik Csapak]

on slower ceph clusters, the write pattern of the ovmf booting process
slows down the boot of the vm, so we turn on caching by default

it seems no other storage (until now) behaves like this. if it does in
the future, we can still add them too, or add a 'cache' property for
the efidisk

Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
---
 PVE/QemuServer.pm | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
index d77981c..96375c3 100644
--- a/PVE/QemuServer.pm
+++ b/PVE/QemuServer.pm
@@ -3307,8 +3307,16 @@ sub config_to_command {
 	    $size_str = ",size=" . (-s $ovmf_vars);
 	}
 
+	# on slower ceph clusters, booting without cache on efidisk can take a while.
+	# enable cache by default, should not be harmful since it's only the efidisk
+	# see #3329
+	my $cache = "";
+	if ($path =~ m/^rbd:/) {
+	    $cache = ',cache=writeback';
+	}
+
 	push @$cmd, '-drive', "if=pflash,unit=0,format=raw,readonly=on,file=$ovmf_code";
-	push @$cmd, '-drive', "if=pflash,unit=1,format=$format,id=drive-efidisk0$size_str,file=$path";
+	push @$cmd, '-drive', "if=pflash,unit=1$cache,format=$format,id=drive-efidisk0$size_str,file=$path";
     }
 
     # load q35 config
-- 
2.20.1

Re: [pve-devel] [PATCH qemu-server] fix #3329: turn on cache=writeback for efidisks on rbd

[Maximilian Hill]

[-- Attachment #1: Type: text/plain, Size: 734 bytes --]

Hello,

On Wed, Jun 16, 2021 at 03:09:33PM +0200, Dominik Csapak wrote:
> on slower ceph clusters, the write pattern of the ovmf booting process
> slows down the boot of the vm, so we turn on caching by default
> 
> +	my $cache = "";
> +	if ($path =~ m/^rbd:/) {
> +	    $cache = ',cache=writeback';
does this work, evven if the rbd storage is not called "rbd"?
> +	}
> +
>  	push @$cmd, '-drive', "if=pflash,unit=0,format=raw,readonly=on,file=$ovmf_code";
> -	push @$cmd, '-drive', "if=pflash,unit=1,format=$format,id=drive-efidisk0$size_str,file=$path";
> +	push @$cmd, '-drive', "if=pflash,unit=1$cache,format=$format,id=drive-efidisk0$size_str,file=$path";
>      }
>  
>      # load q35 config
> -- 
> 2.20.1

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [pve-devel] [PATCH qemu-server] fix #3329: turn on cache=writeback for efidisks on rbd

[Thomas Lamprecht]

hi,

On 16.06.21 15:25, Maximilian Hill wrote:
> On Wed, Jun 16, 2021 at 03:09:33PM +0200, Dominik Csapak wrote:
>> on slower ceph clusters, the write pattern of the ovmf booting process
>> slows down the boot of the vm, so we turn on caching by default
>>
>> +	my $cache = "";
>> +	if ($path =~ m/^rbd:/) {
>> +	    $cache = ',cache=writeback';
> does this work, evven if the rbd storage is not called "rbd"?

yes, this has nothing to do with the storage name, the path is already translated here
and "rbd:" tells QEMU to use librdb for accessing this image.

[pve-devel] applied: [PATCH qemu-server] fix #3329: turn on cache=writeback for efidisks on rbd

[Thomas Lamprecht]

On 16.06.21 15:09, Dominik Csapak wrote:
> on slower ceph clusters, the write pattern of the ovmf booting process
> slows down the boot of the vm, so we turn on caching by default
> 
> it seems no other storage (until now) behaves like this. if it does in
> the future, we can still add them too, or add a 'cache' property for
> the efidisk
> 
> Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
> ---
>  PVE/QemuServer.pm | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
>

applied, thanks!

I added a cfg2cmd test and made it a shorter ternary operation as followup