all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: "Michael Köppl" <m.koeppl@proxmox.com>
To: Proxmox VE development discussion <pve-devel@lists.proxmox.com>,
	Daniel Kral <d.kral@proxmox.com>
Subject: Re: [pve-devel] [PATCH ha-manager v3 04/13] rules: add global checks between node and resource affinity rules
Date: Tue, 29 Jul 2025 13:44:17 +0200	[thread overview]
Message-ID: <f2305d5b-5d19-4c2f-937c-5ca7977c73d7@proxmox.com> (raw)
In-Reply-To: <20250704182102.467624-5-d.kral@proxmox.com>

On 7/4/25 20:20, Daniel Kral wrote:
> 
> diff --git a/src/PVE/HA/Rules.pm b/src/PVE/HA/Rules.pm
> index 3121424..892e7aa 100644
> --- a/src/PVE/HA/Rules.pm
> +++ b/src/PVE/HA/Rules.pm
> @@ -6,6 +6,7 @@ use warnings;
>  use PVE::JSONSchema qw(get_standard_option);
>  use PVE::Tools;
>  
> +use PVE::HA::HashTools qw(set_intersect set_union sets_are_disjoint);
>  use PVE::HA::Tools;
>  
>  use base qw(PVE::SectionConfig);
> @@ -476,4 +477,197 @@ sub get_next_ordinal : prototype($) {
>      return $current_order + 1;
>  }
>  
> +=head1 INTER-PLUGIN RULE CHECKERS
> +
> +=cut
> +
> +=head3 check_single_priority_node_affinity_in_resource_affinity_rules(...)
> +
> +Returns a list of resource affinity rule ids, defined in
> +C<$resource_affinity_rules>, where the resources in the resource affinity rule
> +are in node affinity rules, defined in C<$node_affinity_rules>, which have
> +multiple priority groups defined.
> +
> +That is, the resource affinity rule cannot be statically checked to be feasible
> +as the selection of the priority group is dependent on the currently online
> +nodes.
> +
> +If there are none, the returned list is empty.
> +
> +=cut
> +
> +sub check_single_priority_node_affinity_in_resource_affinity_rules {
> +    my ($resource_affinity_rules, $node_affinity_rules) = @_;
> +
> +    my @errors = ();
> +
> +    while (my ($resource_affinity_id, $resource_affinity_rule) = each %$resource_affinity_rules) {
> +        my $priority;
> +        my $resources = $resource_affinity_rule->{resources};
> +
> +        for my $node_affinity_id (keys %$node_affinity_rules) {
> +            my $node_affinity_rule = $node_affinity_rules->{$node_affinity_id};
> +
> +            next if sets_are_disjoint($resources, $node_affinity_rule->{resources});

Just to note this, since we had just discussed it off-list:
I noted during my review of v2 that I would expect a resource affinity
rule to only fail this rule check if its node affinity rule contained
more than 1 node [0].

One option would be to add an explicit check for the length of the list
of nodes in the node affinity rule of the given resource. If it is 1,
move on to the next iteration. This results in a conflict for the
resource affinity rule because of multiple priorities if there is either
one rule with > 1 nodes with at least one priority assigned (case A) or
if there are multiple node affinity rules for the same resource with a
total of > 1 nodes and at least one priority assigned across the rules
(case B). However, since case B also means that the node affinity rules
are in conflict and therefore not applied, this should probably not
result in a conflict for the resource affinity rule.

The other option, as Dano suggested, would be to instead move the
$priority variable to the inner loop:

    while (my ($resource_affinity_id, $resource_affinity_rule) = each
%$resource_affinity_rules) {
        my $resources = $resource_affinity_rule->{resources};

        for my $node_affinity_id (keys %$node_affinity_rules) {
            my $node_affinity_rule =
$node_affinity_rules->{$node_affinity_id};
            my $priority;

            next if sets_are_disjoint($resources,
$node_affinity_rule->{resources});

            for my $node (values %{ $node_affinity_rule->{nodes} }) {
                $priority = $node->{priority} if !defined($priority);

                if ($priority != $node->{priority}) {
                    push @errors, $resource_affinity_id;
                    last; # early return to check next resource affinity
rule
                }
            }
        }
    }

This would result in a conflict for case A, but not for case B, which
seems more in line with the general handling of reconciling resource and
node affinity rules.

[0]
https://lore.proxmox.com/pve-devel/f82ae0b8-0dbe-497e-8cbd-c7c2f6a7a9c6@proxmox.com/

> +
> +            for my $node (values %{ $node_affinity_rule->{nodes} }) {
> +                $priority = $node->{priority} if !defined($priority);
> +
> +                if ($priority != $node->{priority}) {
> +                    push @errors, $resource_affinity_id;
> +                    last; # early return to check next resource affinity rule
> +                }
> +            }
> +        }
> +    }
> +
> +    @errors = sort @errors;
> +    return \@errors;
> +}
> +
> +__PACKAGE__->register_check(
> +    sub {
> +        my ($args) = @_;
> +
> +        return check_single_priority_node_affinity_in_resource_affinity_rules(
> +            $args->{resource_affinity_rules},
> +            $args->{node_affinity_rules},
> +        );
> +    },
> +    sub {
> +        my ($ruleids, $errors) = @_;
> +
> +        for my $ruleid (@$ruleids) {
> +            push @{ $errors->{$ruleid}->{resources} },
> +                "resources are in node affinity rules with multiple priorities";
> +        }
> +    },
> +);
> +
> +=head3 check_positive_resource_affinity_node_affinity_consistency(...)
> +
> +Returns a list of positive resource affinity rule ids, defined in
> +C<$positive_rules>, where the resources in the positive resource affinity rule
> +are restricted to a disjoint set of nodes by their node affinity rules, defined
> +in C<$node_affinity_rules>.
> +
> +That is, the positive resource affinity rule cannot be fullfilled as the
> +resources cannot be placed on the same node.
> +
> +If there are none, the returned list is empty.
> +
> +=cut
> +
> +sub check_positive_resource_affinity_node_affinity_consistency {
> +    my ($positive_rules, $node_affinity_rules) = @_;
> +
> +    my @errors = ();
> +
> +    while (my ($positiveid, $positive_rule) = each %$positive_rules) {
> +        my $allowed_nodes;
> +        my $resources = $positive_rule->{resources};
> +
> +        for my $node_affinity_id (keys %$node_affinity_rules) {
> +            my ($node_affinity_resources, $node_affinity_nodes) =
> +                $node_affinity_rules->{$node_affinity_id}->@{qw(resources nodes)};
> +
> +            next if sets_are_disjoint($resources, $node_affinity_resources);
> +
> +            $allowed_nodes = { $node_affinity_nodes->%* } if !defined($allowed_nodes);
> +            $allowed_nodes = set_intersect($allowed_nodes, $node_affinity_nodes);
> +
> +            if (keys %$allowed_nodes < 1) {
> +                push @errors, $positiveid;
> +                last; # early return to check next positive resource affinity rule
> +            }
> +        }
> +    }
> +
> +    @errors = sort @errors;
> +    return \@errors;
> +}
> +
> +__PACKAGE__->register_check(
> +    sub {
> +        my ($args) = @_;
> +
> +        return check_positive_resource_affinity_node_affinity_consistency(
> +            $args->{positive_rules},
> +            $args->{node_affinity_rules},
> +        );
> +    },
> +    sub {
> +        my ($ruleids, $errors) = @_;
> +
> +        for my $ruleid (@$ruleids) {
> +            push @{ $errors->{$ruleid}->{resources} },
> +                "two or more resources are restricted to different nodes";
> +        }
> +    },
> +);
> +
> +=head3 check_negative_resource_affinity_node_affinity_consistency(...)
> +
> +Returns a list of negative resource affinity rule ids, defined in
> +C<$negative_rules>, where the resources in the negative resource affinity rule
> +are restricted to less nodes than needed to keep them separate by their node
> +affinity rules, defined in C<$node_affinity_rules>.
> +
> +That is, the negative resource affinity rule cannot be fullfilled as there are
> +not enough nodes to spread the resources on.
> +
> +If there are none, the returned list is empty.
> +
> +=cut
> +
> +sub check_negative_resource_affinity_node_affinity_consistency {
> +    my ($negative_rules, $node_affinity_rules) = @_;
> +
> +    my @errors = ();
> +
> +    while (my ($negativeid, $negative_rule) = each %$negative_rules) {
> +        my $allowed_nodes = {};
> +        my $located_resources;
> +        my $resources = $negative_rule->{resources};
> +
> +        for my $node_affinity_id (keys %$node_affinity_rules) {
> +            my ($node_affinity_resources, $node_affinity_nodes) =
> +                $node_affinity_rules->{$node_affinity_id}->@{qw(resources nodes)};
> +            my $common_resources = set_intersect($resources, $node_affinity_resources);
> +
> +            next if keys %$common_resources < 1;
> +
> +            $located_resources = set_union($located_resources, $common_resources);
> +            $allowed_nodes = set_union($allowed_nodes, $node_affinity_nodes);
> +
> +            if (keys %$allowed_nodes < keys %$located_resources) {
> +                push @errors, $negativeid;
> +                last; # early return to check next negative resource affinity rule
> +            }
> +        }
> +    }
> +
> +    @errors = sort @errors;
> +    return \@errors;
> +}
> +
> +__PACKAGE__->register_check(
> +    sub {
> +        my ($args) = @_;
> +
> +        return check_negative_resource_affinity_node_affinity_consistency(
> +            $args->{negative_rules},
> +            $args->{node_affinity_rules},
> +        );
> +    },
> +    sub {
> +        my ($ruleids, $errors) = @_;
> +
> +        for my $ruleid (@$ruleids) {
> +            push @{ $errors->{$ruleid}->{resources} },
> +                "two or more resources are restricted to less nodes than available to the resources";
> +        }
> +    },
> +);
> +
>  1;
> diff --git a/src/PVE/HA/Rules/ResourceAffinity.pm b/src/PVE/HA/Rules/ResourceAffinity.pm
> index 57ccc09..b024c93 100644
> --- a/src/PVE/HA/Rules/ResourceAffinity.pm
> +++ b/src/PVE/HA/Rules/ResourceAffinity.pm
> @@ -167,7 +167,8 @@ __PACKAGE__->register_check(
>          my ($args) = @_;
>  
>          return check_negative_resource_affinity_resources_count(
> -            $args->{negative_rules}, $args->{nodes},
> +            $args->{negative_rules},
> +            $args->{nodes},
>          );
>      },
>      sub {



_______________________________________________
pve-devel mailing list
pve-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel


  reply	other threads:[~2025-07-29 11:43 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-07-04 18:20 [pve-devel] [PATCH container/docs/ha-manager/manager/qemu-server v3 00/19] HA " Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 01/13] rules: introduce plugin-specific canonicalize routines Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 02/13] rules: add haenv node list to the rules' canonicalization stage Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 03/13] rules: introduce resource affinity rule plugin Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 04/13] rules: add global checks between node and resource affinity rules Daniel Kral
2025-07-29 11:44   ` Michael Köppl [this message]
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 05/13] usage: add information about a service's assigned nodes Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 06/13] manager: apply resource affinity rules when selecting service nodes Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 07/13] manager: handle resource affinity rules in manual migrations Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 08/13] sim: resources: add option to limit start and migrate tries to node Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 09/13] test: ha tester: add test cases for negative resource affinity rules Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 10/13] test: ha tester: add test cases for positive " Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 11/13] test: ha tester: add test cases for static scheduler resource affinity Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 12/13] test: rules: add test cases for resource affinity rules Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH ha-manager v3 13/13] api: resources: add check for resource affinity in resource migrations Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH docs v3 1/1] ha: add documentation about ha resource affinity rules Daniel Kral
2025-07-08 16:08   ` Shannon Sterz
2025-07-09  6:19     ` Friedrich Weber
2025-07-30 10:05     ` Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH manager v3 1/3] ui: ha: rules: add " Daniel Kral
2025-07-04 18:20 ` [pve-devel] [PATCH manager v3 2/3] ui: migrate: lxc: display precondition messages for ha resource affinity Daniel Kral
2025-07-04 18:21 ` [pve-devel] [PATCH manager v3 3/3] ui: migrate: vm: " Daniel Kral
2025-07-04 18:21 ` [pve-devel] [PATCH container v3 1/1] api: introduce migration preconditions api endpoint Daniel Kral
2025-07-04 18:21 ` [pve-devel] [PATCH qemu-server v3 1/1] api: migration preconditions: add checks for ha resource affinity rules Daniel Kral

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f2305d5b-5d19-4c2f-937c-5ca7977c73d7@proxmox.com \
    --to=m.koeppl@proxmox.com \
    --cc=d.kral@proxmox.com \
    --cc=pve-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal