public inbox for pve-devel@lists.proxmox.com
 help / color / mirror / Atom feed
From: Daniel Kral <d.kral@proxmox.com>
To: pve-devel@lists.proxmox.com
Subject: Re: [pve-devel] [PATCH ha-manager v2 06/26] rules: add global checks between location and colocation rules
Date: Tue, 1 Jul 2025 13:02:52 +0200	[thread overview]
Message-ID: <7eeb83ea-d5d1-4fce-956d-414bf7c446b0@proxmox.com> (raw)
In-Reply-To: <20250620143148.218469-11-d.kral@proxmox.com>

On 6/20/25 16:31, Daniel Kral wrote:
> +=head3 check_positive_colocation_location_consistency($positive_rules, $location_rules)
> +
> +Returns a list of positive colocation rule ids defined in C<$positive_rules>,
> +where the services in the positive colocation rule are restricted to a disjoint
> +set of nodes by their location rules, defined in C<$location_rules>. That is,
> +the positive colocation rule cannot be fullfilled as the services cannot be
> +placed on the same node.
> +
> +If there are none, the returned list is empty.
> +
> +=cut
> +
> +sub check_positive_colocation_location_consistency {
> +    my ($positive_rules, $location_rules) = @_;
> +
> +    my @errors = ();
> +
> +    while (my ($positiveid, $positive_rule) = each %$positive_rules) {
> +        my $allowed_nodes;
> +        my $services = $positive_rule->{services};
> +
> +        for my $locationid (keys %$location_rules) {
> +            my $location_rule = $location_rules->{$locationid};
> +
> +            next if !$location_rule->{strict};

The "strict" requirement will be removed in a v3.

Service affinity (colocation rules) is determined after node affinity 
(location rules). That is, service affinity selects from the nodes, 
which are in the highest priority node group determined by the node 
affinity rules. Even if a service is in a non-strict node affinity rule, 
the service affinity can only select one of the highest priority nodes, 
no matter if it is strict or non-strict...

I also realized now that it is still in question what positive service 
affinity rules should be allowed if one (or more) of their services are 
also in a node affinity rule.

Consider the case where vm:101, vm:102 and vm:103 must be kept on the 
same node and only one is in a node affinity rule restricting vm:101 to 
only node1. What does that node affinity rule state? Should that rule 
infer that vm:102 and vm:103 are also in the node affinity rule now 
(rather implicit for the user)?

I'd rather make these combinations invalid and remind the user that all 
services should be put in the node affinity rule first with the same 
node selection and then they can create the service affinity rule, but 
feedback on that would be much appreciated.

> +            next if PVE::HashTools::sets_are_disjoint($services, $location_rule->{services});
> +
> +            $allowed_nodes = { $location_rule->{nodes}->%* } if !defined($allowed_nodes);
> +            $allowed_nodes = PVE::HashTools::set_intersect($allowed_nodes, $location_rule->{nodes});
> +
> +            if (keys %$allowed_nodes < 1) {
> +                push @errors, $positiveid;
> +                last; # early return to check next positive colocation rule
> +            }
> +        }
> +    }
> +
> +    @errors = sort @errors;
> +    return \@errors;
> +}
> +
> +__PACKAGE__->register_check(
> +    sub {
> +        my ($args) = @_;
> +
> +        return check_positive_colocation_location_consistency(
> +            $args->{positive_rules},
> +            $args->{location_rules},
> +        );
> +    },
> +    sub {
> +        my ($ruleids, $errors) = @_;
> +
> +        for my $ruleid (@$ruleids) {
> +            push @{ $errors->{$ruleid}->{services} },
> +                "two or more services are restricted to different nodes";
> +        }
> +    },
> +);
> +
> +=head3 check_negative_colocation_location_consistency($negative_rules, $location_rules)
> +
> +Returns a list of negative colocation rule ids defined in C<$negative_rules>,
> +where the services in the negative colocation rule are restricted to less nodes
> +than needed to keep them separate by their location rules, defined in
> +C<$location_rules>. That is, the negative colocation rule cannot be fullfilled
> +as there are not enough nodes to spread the services on.
> +
> +If there are none, the returned list is empty.
> +
> +=cut
> +
> +sub check_negative_colocation_location_consistency {
> +    my ($negative_rules, $location_rules) = @_;
> +
> +    my @errors = ();
> +
> +    while (my ($negativeid, $negative_rule) = each %$negative_rules) {
> +        my $allowed_nodes = {};
> +        my $located_services;
> +        my $services = $negative_rule->{services};
> +
> +        for my $locationid (keys %$location_rules) {
> +            my $location_rule = $location_rules->{$locationid};
> +
> +            my $location_services = $location_rule->{services};
> +            my $common_services = PVE::HashTools::set_intersect($services, $location_services);
> +
> +            next if !$location_rule->{strict};

Same argument as above regarding the strictness.

> +            next if keys %$common_services < 1;
> +
> +            $located_services = PVE::HashTools::set_union($located_services, $common_services);
> +            $allowed_nodes = PVE::HashTools::set_union($allowed_nodes, $location_rule->{nodes});
> +
> +            if (keys %$allowed_nodes < keys %$located_services) {
> +                push @errors, $negativeid;
> +                last; # early return to check next negative colocation rule
> +            }
> +        }
> +    }
> +
> +    @errors = sort @errors;
> +    return \@errors;
> +}


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


  reply	other threads:[~2025-07-01 11:02 UTC|newest]

Thread overview: 70+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-06-20 14:31 [pve-devel] [RFC common/cluster/ha-manager/docs/manager v2 00/40] HA " Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH common v2 1/1] introduce HashTools module Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH cluster v2 1/3] cfs: add 'ha/rules.cfg' to observed files Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH cluster v2 2/3] datacenter config: make pve-ha-shutdown-policy optional Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH cluster v2 3/3] datacenter config: introduce feature flag for location rules Daniel Kral
2025-06-23 15:58   ` Thomas Lamprecht
2025-06-24  7:29     ` Daniel Kral
2025-06-24  7:51       ` Thomas Lamprecht
2025-06-24  8:19         ` Daniel Kral
2025-06-24  8:25           ` Thomas Lamprecht
2025-06-24  8:52             ` Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 01/26] tree-wide: make arguments for select_service_node explicit Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 02/26] manager: improve signature of select_service_node Daniel Kral
2025-06-23 16:21   ` Thomas Lamprecht
2025-06-24  8:06     ` Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 03/26] introduce rules base plugin Daniel Kral
2025-07-04 14:18   ` Michael Köppl
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 04/26] rules: introduce location rule plugin Daniel Kral
2025-06-20 16:17   ` Jillian Morgan
2025-06-20 16:30     ` Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 05/26] rules: introduce colocation " Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 06/26] rules: add global checks between location and colocation rules Daniel Kral
2025-07-01 11:02   ` Daniel Kral [this message]
2025-07-04 14:43   ` Michael Köppl
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 07/26] config, env, hw: add rules read and parse methods Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 08/26] manager: read and update rules config Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 09/26] test: ha tester: add test cases for future location rules Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 10/26] resources: introduce failback property in service config Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 11/26] manager: migrate ha groups to location rules in-memory Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 12/26] manager: apply location rules when selecting service nodes Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 13/26] usage: add information about a service's assigned nodes Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 14/26] manager: apply colocation rules when selecting service nodes Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 15/26] manager: handle migrations for colocated services Daniel Kral
2025-06-27  9:10   ` Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 16/26] sim: resources: add option to limit start and migrate tries to node Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 17/26] test: ha tester: add test cases for strict negative colocation rules Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 18/26] test: ha tester: add test cases for strict positive " Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 19/26] test: ha tester: add test cases in more complex scenarios Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 20/26] test: add test cases for rules config Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 21/26] manager: handle negative colocations with too many services Daniel Kral
2025-07-01 12:11   ` Michael Köppl
2025-07-01 12:23     ` Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 22/26] config: prune services from rules if services are deleted from config Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 23/26] api: introduce ha rules api endpoints Daniel Kral
2025-07-04 14:16   ` Michael Köppl
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 24/26] cli: expose ha rules api endpoints to ha-manager cli Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 25/26] api: groups, services: assert use-location-rules feature flag Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH ha-manager v2 26/26] api: services: check for colocations for service motions Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH docs v2 1/5] ha: config: add section about ha rules Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH docs v2 2/5] update static files to include ha rules api endpoints Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH docs v2 3/5] update static files to include use-location-rules feature flag Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH docs v2 4/5] update static files to include ha resources failback flag Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH docs v2 5/5] update static files to include ha service motion return value schema Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH manager v2 1/5] api: ha: add ha rules api endpoints Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH manager v2 2/5] ui: add use-location-rules feature flag Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH manager v2 3/5] ui: ha: hide ha groups if use-location-rules is enabled Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH manager v2 4/5] ui: ha: adapt resources components " Daniel Kral
2025-06-20 14:31 ` [pve-devel] [PATCH manager v2 5/5] ui: ha: add ha rules components and menu entry Daniel Kral
2025-06-30 15:09   ` Michael Köppl
2025-07-01 14:38   ` Michael Köppl
2025-06-20 15:43 ` [pve-devel] [RFC common/cluster/ha-manager/docs/manager v2 00/40] HA colocation rules Daniel Kral
2025-06-20 17:11   ` Jillian Morgan
2025-06-20 17:45     ` DERUMIER, Alexandre via pve-devel
     [not found]     ` <476c41123dced9d560dfbf27640ef8705fd90f11.camel@groupe-cyllene.com>
2025-06-23 15:36       ` Thomas Lamprecht
2025-06-24  8:48         ` Daniel Kral
2025-06-27 12:23           ` Friedrich Weber
2025-06-27 12:41             ` Daniel Kral
2025-06-23  8:11 ` DERUMIER, Alexandre via pve-devel
     [not found] ` <bf973ec4e8c52a10535ed35ad64bf0ec8d1ad37d.camel@groupe-cyllene.com>
2025-06-23 15:28   ` Thomas Lamprecht
2025-06-23 23:21     ` DERUMIER, Alexandre via pve-devel

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=7eeb83ea-d5d1-4fce-956d-414bf7c446b0@proxmox.com \
    --to=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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal