Re: [pve-devel] [PATCH-SERIES container/ha-manager/manager/qemu-server 00/12] HA node affinity blockers (#1497)

[Fiona Ebner <f.ebner@proxmox.com>]

Am 15.12.25 um 4:54 PM schrieb Daniel Kral:
> This patch series implements node affinity rule migration blockers
> similar to the blockers introduced with resource affinity rules.
> 
> 
> The node affinity rule migraton blockers prevent users from migrating HA
> resources to nodes, which would make them migrate somewhere else
> immediately afterwards. This includes:
> 
> - online nodes, which are not part of the strict node affinity rule's
>   allowed node set at all, or
> 
> - online nodes, which are not in the currently highest priority group of
>   the strict or non-strict node affinity rule and the HA resource has
>   failback set.
> 
> 
> The first few patches are some overall cleanup for things the series
> touches + deduplicating the resource_motion_info logic and sharing it
> between the Manager and the public
> PVE::HA::Config::get_resource_motion_info(...), as well as exposing these
> in the relevant VM/LXC API handlers and web interface.
> 
> 
> Since the 'cause' property in 'blocking-resources' is an enum,
> qemu-server and pve-container need version bumps for pve-ha-manager as
> else the API handlers would result in a schema result verification error.
> 
> If we can spare some coupling between those, we could generate the schema
> for these API handlers from the pve-ha-manager package, so future
> versioned bumps are not needed anymore, but I was a bit hesitant to
> implement that in the v1.

There are some nits (see individual patches), but consider the series:

Reviewed-by: Fiona Ebner <f.ebner@proxmox.com>


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