Thanks for reporting this. I closed the original, so we can keep this.

This part should not have a recent change. cc @ChanYiLin follow up.

I ran into this issue today ... is this still on the roadmap?

I ran into this today as well with a fresh Helm install; I'm setting defaultSettings.taintToleration: Storage=true:NoSchedule in my values (with Storage=true:NoSchedule as the taint applied to my 3 Longhorn-only nodes) and it doesn't seem to get applied to the engine-image-ei-04c05bf8 or longhorn-csi-plugin DaemonSets.

defaultSettings:
  createDefaultDiskLabeledNodes: true
  defaultReplicaCount: 2
  taintToleration: Storage=true:NoSchedule
  removeSnapshotsDuringFilesystemTrim: enabled

I can see it clear as day in my longhorn-default-setting ConfigMap though, for what it's worth, but I don't think Longhorn actually picks up this config map; if I browse to Longhorn UI's settings, all those have their default values applied (including an empty toleration in the Danger Zone) and if I add a toleration there, it works. All of the Setting objects in my longhorn-system namespace all have their default values. Here's my ConfigMap:

create-default-disk-labeled-nodes: true

default-replica-count: 2

taint-toleration: Storage=true:NoSchedule

priority-class: longhorn-critical

disable-revision-counter: true

remove-snapshots-during-filesystem-trim: enabled

Happy to help troubleshoot, I'm on v1.30.4+rke2r1. One would think I've hit #2562, but I'm using only settings that are defined in the v1.7.0 values.yaml.

• Longhorn version: v1.7.0
• Installation method: Helm
• Kubernetes distro and version: v1.30.4+rke2r1
• Number of management nodes in the cluster: 3
• Number of worker nodes in the cluster: 6
• Node Configuration:
  • OS type and version: Debian 12 Bookworm
  • CPU per node: 8
  • Memory per node: workers 18 GB, management 8 GB
  • Disk type: 128 GB
• Network bandwidth between the nodes: 1 Gbps
• Underlying Infrastructure: Proxmox
• Number of Longhorn volumes in the cluster: 0

Ah-hah: it was my Helm value for removeSnapshotsDuringFilesystemTrim causing this problem; it needed to be true or false, but not enabled (which is an acceptable value for the persistence section of the values chart, but not the defaultSettings section). I discovered this when writing up a loop of Ansible patch operations to modify the Setting objects after an initial install. With the updated section in my values:

defaultSettings:
  createDefaultDiskLabeledNodes: true
  defaultReplicaCount: '2'
  taintToleration: Storage=true:NoSchedule
  removeSnapshotsDuringFilesystemTrim: true # Note this line!

…my Setting objects are now properly populated, including my taint-toleration Setting.

NAME                                                              VALUE                                         APPLIED   AGE
create-default-disk-labeled-nodes                                 true                                          true      5m13s
default-replica-count                                             2                                             true      5m13s
taint-toleration                                                  Storage=true:NoSchedule                       true      5m13s
remove-snapshots-during-filesystem-trim                           true                                          true      5m12s

So to those having this issue if you're installing from Helm: check the values you have set for all of your defaultSettings keys in your Helm chart; some of them may not map to acceptable Longhorn values or value types.

To the Longhorn maintainers: since these defaultSettings values are picked up and applied the way they are, each key could use additional specific documentation about acceptable values. Many do, but not all!

@mantissahz will it be fixed?
This issue is up already for one year :D

I can only see the issue when doing helm upgrade.
but with a fresh helm install, the tolerations are added as expected.