From mboxrd@z Thu Jan 1 00:00:00 1970
Return-Path: <f.ebner@proxmox.com>
Received: from firstgate.proxmox.com (firstgate.proxmox.com [212.224.123.68])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)
key-exchange X25519 server-signature RSA-PSS (2048 bits))
(No client certificate requested)
by lists.proxmox.com (Postfix) with ESMTPS id 7DA4996490
for <pve-devel@lists.proxmox.com>; Tue, 24 Jan 2023 14:05:21 +0100 (CET)
Received: from firstgate.proxmox.com (localhost [127.0.0.1])
by firstgate.proxmox.com (Proxmox) with ESMTP id 63D0A3B86
for <pve-devel@lists.proxmox.com>; Tue, 24 Jan 2023 14:04:51 +0100 (CET)
Received: from proxmox-new.maurer-it.com (proxmox-new.maurer-it.com
[94.136.29.106])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)
key-exchange X25519 server-signature RSA-PSS (2048 bits))
(No client certificate requested)
by firstgate.proxmox.com (Proxmox) with ESMTPS
for <pve-devel@lists.proxmox.com>; Tue, 24 Jan 2023 14:04:50 +0100 (CET)
Received: from proxmox-new.maurer-it.com (localhost.localdomain [127.0.0.1])
by proxmox-new.maurer-it.com (Proxmox) with ESMTP id 15AA245EAA;
Tue, 24 Jan 2023 14:04:50 +0100 (CET)
Message-ID: <68928d13-7bd1-90bc-88e7-5db7f727637a@proxmox.com>
Date: Tue, 24 Jan 2023 14:04:48 +0100
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
Thunderbird/102.5.0
From: Fiona Ebner <f.ebner@proxmox.com>
To: pve-devel@lists.proxmox.com, "aderumier@odiso.com" <aderumier@odiso.com>
References: <20230104064303.2898194-1-aderumier@odiso.com>
<20230104064303.2898194-3-aderumier@odiso.com>
Content-Language: en-US
In-Reply-To: <20230104064303.2898194-3-aderumier@odiso.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
X-SPAM-LEVEL: Spam detection results: 0
AWL 0.594 Adjusted score from AWL reputation of From: address
BAYES_00 -1.9 Bayes spam probability is 0 to 1%
KAM_DMARC_STATUS 0.01 Test Rule for DKIM or SPF Failure with Strict Alignment
NICE_REPLY_A -1.148 Looks like a legit reply (A)
SPF_HELO_NONE 0.001 SPF: HELO does not publish an SPF Record
SPF_PASS -0.001 SPF: sender matches SPF record
URIBL_BLOCKED 0.001 ADMINISTRATOR NOTICE: The query to URIBL was blocked. See
http://wiki.apache.org/spamassassin/DnsBlocklists#dnsbl-block for more
information. [qemuconfig.pm, memory.pm, qemumigrate.pm, qemu.pm, qemuserver.pm,
helpers.pm]
Subject: Re: [pve-devel] [PATCH v2 qemu-server 2/9] add memory parser
X-BeenThere: pve-devel@lists.proxmox.com
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Proxmox VE development discussion <pve-devel.lists.proxmox.com>
List-Unsubscribe: <https://lists.proxmox.com/cgi-bin/mailman/options/pve-devel>,
<mailto:pve-devel-request@lists.proxmox.com?subject=unsubscribe>
List-Archive: <http://lists.proxmox.com/pipermail/pve-devel/>
List-Post: <mailto:pve-devel@lists.proxmox.com>
List-Help: <mailto:pve-devel-request@lists.proxmox.com?subject=help>
List-Subscribe: <https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel>,
<mailto:pve-devel-request@lists.proxmox.com?subject=subscribe>
X-List-Received-Date: Tue, 24 Jan 2023 13:05:21 -0000
Am 04.01.23 um 07:42 schrieb Alexandre Derumier:
> Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
> ---
> PVE/API2/Qemu.pm | 12 ++++++-
> PVE/QemuConfig.pm | 4 +--
> PVE/QemuMigrate.pm | 6 ++--
> PVE/QemuServer.pm | 24 ++++++-------
> PVE/QemuServer/Helpers.pm | 3 +-
> PVE/QemuServer/Memory.pm | 74 ++++++++++++++++++++++++++++++++-------
> 6 files changed, 91 insertions(+), 32 deletions(-)
>
> diff --git a/PVE/API2/Qemu.pm b/PVE/API2/Qemu.pm
> index c87602d..4ffa973 100644
> --- a/PVE/API2/Qemu.pm
> +++ b/PVE/API2/Qemu.pm
> @@ -32,6 +32,7 @@ use PVE::QemuServer::Drive;
> use PVE::QemuServer::ImportDisk;
> use PVE::QemuServer::Monitor qw(mon_cmd);
> use PVE::QemuServer::Machine;
> +use PVE::QemuServer::Memory qw(get_current_memory);
> use PVE::QemuMigrate;
> use PVE::RPCEnvironment;
> use PVE::AccessControl;
> @@ -1608,7 +1609,16 @@ my $update_vm_api = sub {
> }
>
> if ($param->{memory} || defined($param->{balloon})) {
> - my $maxmem = $param->{memory} || $conf->{pending}->{memory} || $conf->{memory} || $defaults->{memory};
Nit: seems like this was the only usage of $defaults in the whole
function, so the variable could be removed too
> +
> + my $maxmem = undef;
> + if ($param->{memory}) {
> + $maxmem = get_current_memory($param->{memory});
> + } elsif ($conf->{pending}->{memory}) {
> + $maxmem = get_current_memory($conf->{pending}->{memory});
> + } else {
> + $maxmem = get_current_memory($conf->{memory});
> + }
> +
> my $balloon = defined($param->{balloon}) ? $param->{balloon} : $conf->{pending}->{balloon} || $conf->{balloon};
>
> die "balloon value too large (must be smaller than assigned memory)\n"
(...)
> diff --git a/PVE/QemuServer.pm b/PVE/QemuServer.pm
> index 39fc6b0..5847a78 100644
> --- a/PVE/QemuServer.pm
> +++ b/PVE/QemuServer.pm
(...)
> @@ -5113,7 +5110,7 @@ sub vmconfig_hotplug_pending {
> $vmid, $opt, $value, $arch, $machine_type);
> } elsif ($opt =~ m/^memory$/) { #dimms
> die "skip\n" if !$hotplug_features->{memory};
> - $value = PVE::QemuServer::Memory::qemu_memory_hotplug($vmid, $conf, $defaults, $opt, $value);
> + PVE::QemuServer::Memory::qemu_memory_hotplug($vmid, $conf, $defaults, $conf->{pending}->{$opt});
By removing the assignment to $value here, the passed-in value will be
written to the config below, which does not reflect reality. This is a
breaking change:
root@pve701 ~ # qm set 131 --memory 1024
update VM 131: -memory 1024
root@pve701 ~ # cat /etc/pve/qemu-server/131.conf | grep memory:
memory: 1024
root@pve701 ~ # qm start 131
root@pve701 ~ # qm set 131 --memory 1025
update VM 131: -memory 1025
root@pve701 ~ # cat /etc/pve/qemu-server/131.conf | grep memory:
memory: 1025
But it really is 1536, because a 512M dimm was added. And now, because
the value is wrong, the following (or with any higher value) fails:
root@pve701 ~ # qm set 131 --memory 1026
update VM 131: -memory 1026
400 Parameter verification failed.
memory: hotplug problem - VM 131 qmp command 'object-add' failed -
attempt to add duplicate property 'mem-dimm0' to object (type 'container')
qm set <vmid> [OPTIONS]
While the current behavior is arguably also a bit surprising, it
reflects the actual values, and doesn't fail:
root@pve701 ~ # qm set 131 --memory 1024
update VM 131: -memory 1024
root@pve701 ~ # cat /etc/pve/qemu-server/131.conf | grep memory:
memory: 1024
root@pve701 ~ # qm start 131
root@pve701 ~ # qm set 131 --memory 1025
update VM 131: -memory 1025
root@pve701 ~ # cat /etc/pve/qemu-server/131.conf | grep memory:
memory: 1536
root@pve701 ~ # qm set 131 --memory 1026
update VM 131: -memory 1026
try to unplug memory dimm dimm0
root@pve701 ~ # cat /etc/pve/qemu-server/131.conf | grep memory:
memory: 1024
Why not keep assigning and returning the value from
qemu_memory_hotplug() in the cases it's not already written there?
> } elsif ($opt eq 'cpuunits') {
> my $new_cpuunits = PVE::CGroup::clamp_cpu_shares($conf->{pending}->{$opt}); #clamp
> $cgroup->change_cpu_shares($new_cpuunits);
(...)
> diff --git a/PVE/QemuServer/Memory.pm b/PVE/QemuServer/Memory.pm
> index 6c1cd94..59e51c8 100644
> --- a/PVE/QemuServer/Memory.pm
> +++ b/PVE/QemuServer/Memory.pm
> @@ -8,10 +8,45 @@ use PVE::Exception qw(raise raise_param_exc);
>
> use PVE::QemuServer;
> use PVE::QemuServer::Monitor qw(mon_cmd);
> +use base qw(Exporter);
> +
> +our @EXPORT_OK = qw(
> +get_current_memory
> +);
>
> my $MAX_NUMA = 8;
> my $STATICMEM = 1024;
>
> +my $memory_fmt = {
Since QemuServer.pm references it, shouldn't it be 'our'?
> + current => {
> + description => "Current amount of online RAM for the VM in MB. This is the maximum available memory when"
> + ." you use the balloon device.",
> + type => 'integer',
> + default_key => 1,
> + optional => 1,
Why optional?
> + minimum => 16,
> + default => 512,
> + }
> +};
> +
> +sub print_memory {
> + my $memory = shift;
> +
> + return PVE::JSONSchema::print_property_string($memory, $memory_fmt);
> +}
> +
> +sub parse_memory {
> + my ($value) = @_;
> +
> + my $current_default = $memory_fmt->{current}->{default};
> + my $res = { current => $current_default };
> + return $res if !defined($value);
Style nit: Could be one line:
return { current => $memory_fmt->{current}->{default} } if !defined($value).
> +
> + $res = eval { PVE::JSONSchema::parse_property_string($memory_fmt, $value) };
> + die $@ if $@;
Style nit: No need for eval+die if the error is just passed on as-is
> + return $res;
> +}
> +
> my $_host_bits;
> my sub get_host_phys_address_bits {
> return $_host_bits if defined($_host_bits);
> @@ -63,6 +98,13 @@ my sub get_max_mem {
> return $bits_to_max_mem > 4*1024*1024 ? 4*1024*1024 : $bits_to_max_mem;
> }
>
> +sub get_current_memory {
> + my ($value) = @_;
> +
> + my $memory = parse_memory($value);
> + return $memory->{current};
> +}
> +
> sub get_numa_node_list {
> my ($conf) = @_;
> my @numa_map;
> @@ -155,17 +197,21 @@ sub foreach_reverse_dimm {
> }
>
> sub qemu_memory_hotplug {
> - my ($vmid, $conf, $defaults, $opt, $value) = @_;
> + my ($vmid, $conf, $defaults, $value) = @_;
Removing the unused $opt from qemu_memory_hotplug() could also be its
own patch.
> +
> + return if !PVE::QemuServer::check_running($vmid);
>
> - return $value if !PVE::QemuServer::check_running($vmid);
> + my $oldmem = parse_memory($conf->{memory});
> + my $newmem = parse_memory($value);
> +
> + my $memory = $oldmem->{current};
> + $value = $newmem->{current};
> +
> + return if $value == $memory;
>
> my $sockets = 1;
> $sockets = $conf->{sockets} if $conf->{sockets};
>
> - my $memory = $conf->{memory} || $defaults->{memory};
> - $value = $defaults->{memory} if !$value;
> - return $value if $value == $memory;
> -
> my $static_memory = $STATICMEM;
> $static_memory = $static_memory * $sockets if ($conf->{hugepages} && $conf->{hugepages} == 1024);
>
(...)
> @@ -270,7 +319,8 @@ sub qemu_dimm_list {
> sub config {
> my ($conf, $vmid, $sockets, $cores, $defaults, $hotplug_features, $cmd) = @_;>
The $defaults parameter can be dropped now
> - my $memory = $conf->{memory} || $defaults->{memory};
> + my $memory = get_current_memory($conf->{memory});
> +
> my $static_memory = 0;
>
> if ($hotplug_features->{memory}) {
(...)