Bug: anti-affinity not working

[ybizeul]

General

plb_exclude_something set, but two VMs are still on same node.
Logs indicates that rebalance is needed, but not performed (k8s-prod-worker-kpd8v)

Weighting

Score: 10

Config

[proxmox]
api_host: 10.65.179.101
api_user: root@pam
api_pass: ******
verify_ssl: 0
[vm_balancing]
enable: 1
method: memory
mode: used
maintenance_nodes: dummynode03,dummynode04
ignore_nodes: pve5
ignore_vms: testvm01,testvm02
[storage_balancing]
enable: 0
[update_service]
enable: 0
[api]
enable: 0
[service]
daemon: 1
schedule: 24
log_verbosity: DEBUG
config_version: 3

Log

Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [balanciness-validation]: Rebalancing for memory is not needed. Highest usage: 97% | Lowest usage: 90%.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-vm-calculator]: Balancing calculations done.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-maintenance-vm-calculator]: Maintenance mode for the following hosts defined: set()
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Node pve1 has been added as an already used node to the anti-affinity group plb_exclude_worker.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: No rebalancing for VM k8s-prod-worker-2fk8w needed due to any anti-affinity group policies.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Rebalancing of VM tco-linux is needed due to anti-affinity group policy.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [random-node-getter]: New random node pve1 evaluated for vm tco-linux in run 1.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Node pve2 has been added as an already used node to the anti-affinity group plb_exclude_worker.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: No rebalancing for VM k8s-prod-worker-vpz8c needed due to any anti-affinity group policies.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Rebalancing of VM k8s-prod-worker-kpd8v is needed due to anti-affinity group policy.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [random-node-getter]: New random node pve1 evaluated for vm k8s-prod-worker-kpd8v in run 1.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Node pve3 has been added as an already used node to the anti-affinity group plb_exclude_k8s_control.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: No rebalancing for VM k8s-prod-control-plane-2q7r5 needed due to any anti-affinity group policies.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: Node pve2 has been added as an already used node to the anti-affinity group plb_exclude_k8s_control.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [rebalancing-tags-group-exclude]: No rebalancing for VM k8s-prod-control-plane-vs8tp needed due to any anti-affinity group policies.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [vm-rebalancing-executor]: No rebalancing needed.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [cli-output-generator]: Start rebalancing vms to their new nodes.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [cli-output-generator]: No rebalancing needed.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [post-validations]: All post-validations succeeded.
Dec 01 22:48:12 pve1 proxlb[3729287]:  ProxLB: Info: [daemon]: Running in daemon mode. Next run in 24 hours.

Meta

Version: 1.0.5
Installed from: deb
Running as: local on proxmox

[erth5]

Same Problem on pve-manager/8.3.1/fb48e850ef9dde27 (running kernel: 6.8.12-4-pve)
I exclude a qourum sever
ignore_nodes: pve0
but it migrates VMs to there
Dec 10 16:05:47 pve1 proxlb[8185]: ProxLB: Info: [vm-rebalancing-executor]: Rebalancing VM deb3 from node pve1 to node pve0

installed from dpgk
Version: 1.0.5
api on 127.0.0.1

[laitco]

Same error with the tags of like plb_exclude_something. Rule seeams to be ignored and VMs are not migrated to seperate hosts to keep them apart.

My actual workarround:
plb_ignore_vm and setting a HA group rule to specific host

Linux 6.8.12-8-pve (8.3.4)
Version: ProxLB version 1.0.5

[gyptazy]

This will be fixed with version 1.1.0 in #114