NVMe sample: Fix nvme ids #446
Conversation
OpenShift image has the NVMe `hostid` and `hostnqn` hardcoded, so it ends up being the same in all the control plane nodes, which is problematic. This `MachineConfig` fixes this issue by recreating both files when the `hostid` doesn't match the system-uuid of the machine it is running on. Bugs: - https://issues.redhat.com/browse/OCPBUGS-34629 - openshift/os#1519 - openshift/os#1520
|
/retest |
1 similar comment
|
/retest |
| metadata: | ||
| labels: | ||
| component: fix-nvme-ids | ||
| machineconfiguration.openshift.io/role: master |
There was a problem hiding this comment.
Sorry, I just spotted this now, but should the role be worker? I think our other machineConfig specify master nodes, and that works in our small environments where nodes are both.
There was a problem hiding this comment.
For our samples it should be master, like in all the other MachineConfigs we have in our samples.
It's only worker when we are deploying in a 3 master and N workers deployment and OpenStack control plane will be running on the workers.
This is described in the generic documents PR:
cinder-operator/docs/user-guide/cinder.md
Line 156 in ff49b64
> **⚠ Attention:** `MachineConfig` examples in the following sections are using
label `machineconfiguration.openshift.io/role: worker` instructing OpenShift to
apply those changes on `worker` nodes (`workers` is an automatically created
`MachineConfigPool`) This assumes we have an OpenShift cluster with 3 master
nodes and 3 `worker` nodes. If we are deploying in a 3 node OpenShift cluster
where all nodes are `master` and `worker` we need to change it to use `master`
instead (`master` is another automatically created `MachineConfigPool`).
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: Akrog, ASBishop The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/test cinder-operator-build-deploy-kuttl |
openshift-merge-bot
bot
merged commit a808b25
into
openstack-k8s-operators:main
OpenShift image has the NVMe
hostidandhostnqnhardcoded, so it ends up being the same in all the control plane nodes, which is problematic.This
MachineConfigfixes this issue by recreating both files when thehostiddoesn't match the system-uuid of the machine it is running on.Bugs:
/usr/etc/nvme/hostid/usr/etc/nvme/hostnqnopenshift/os#1519