Hybrid VMs at large scale improvements by afcollins · Pull Request #733 · redhat-performance/jetlag

afcollins · 2025-10-31T20:23:53Z

Enables hugepages on the hypervisor and libvirt VMs.
Fixes some issues with hugetlb pages.

Also, adds interfaces that can be used as virtual functions.

Some changes were generated using Cursor and the claude-4-sonnet model.

docs/troubleshooting.md

ansible/roles/hv-vm-create/templates/kvm-def.xml.j2

ansible/roles/hv-vm-destroy/tasks/main.yml

ansible/roles/hv-vm-start/tasks/main.yml

ansible/roles/wait-hosts-discovered/tasks/set_hostname_role.yml

ansible/roles/ocp-scale-out-csr/tasks/main.yml

ansible/vars/hv.yml

ansible.cfg

afcollins · 2025-11-20T18:24:29Z

Addressed all comments. Reverted some unnecessary changes. Renamed a variable. Added a variable for VFs in VMs. Please take another look when you can.

I still don't have a way to validate these changes for right now. :/ But may be able to do so in early December.

ansible/roles/ocp-scale-out-csr/tasks/check_nodes_joined.yml

ansible/vars/lab.yml

ansible/vars/scale_out.sample.yml

ansible/hv-vm-stop-all.yml

ansible/mno-add-vm-workers.yml

afcollins · 2026-01-30T05:16:29Z

Changing to draft while I split into more atomic PRs.

ansible/roles/hv-vm-create/tasks/main.yml

mcornea · 2026-02-18T10:56:31Z

Hi @afcollins Is this PR ready for merge?

Update the documentation. Signed-off-by: Andrew Collins <ancollin@redhat.com>

Signed-off-by: Andrew Collins <ancollin@redhat.com>

This update enhances the usability and understanding of the deployment process. Co-Authored-By: Cursor agent, claude-4.6-opus model Signed-off-by: Andrew Collins <ancollin@redhat.com>

afcollins · 2026-02-26T18:28:06Z

It is not quite ready yet.

The documentation was never finished, so I filled it out and reduced the complexity.

Rebased on main.

Seems memory isn't being calculated correctly, so I need to dig into it.

VM vm00001 will use 2 1Gi hugepages (2G total)

afcollins · 2026-02-27T01:06:14Z

Figured it out. The vm_hugepage_count variable was used in the debug message.

After further review, the variable is only used for the message and the VM will take as many hugepages as required based on its defined memory. Therefore, I think it makes more sense to remove the variable entirely and only use the memory defined on the hv_vm host entry.

Tested and it works. Should be all good to go now.

TASK [hv-vm-create : Set hugepage count for VM] *******************************
Friday 27 February 2026  01:00:53 +0000 (0:00:00.015)       0:00:01.139 *******
ok: [vm00001]

TASK [hv-vm-create : Check host hugepages availability] ***********************
Friday 27 February 2026  01:00:53 +0000 (0:00:00.029)       0:00:01.169 *******
changed: [vm00001 -> e35-......com]

TASK [hv-vm-create : Validate sufficient hugepages available] *****************
Friday 27 February 2026  01:00:54 +0000 (0:00:00.738)       0:00:01.908 *******
skipping: [vm00001]

TASK [hv-vm-create : Display hugepages configuration for VM] ******************
Friday 27 February 2026  01:00:54 +0000 (0:00:00.026)       0:00:01.934 *******
ok: [vm00001] => {
    "msg": "VM vm00001 will use 18 1Gi hugepages (18G total)"
}

mcornea · 2026-02-27T10:52:16Z

Awesome, thanks @afcollins !

mcornea · 2026-02-27T10:52:19Z

/lgtm

mcornea · 2026-02-27T10:52:22Z

/approve

openshift-ci · 2026-02-27T10:52:28Z

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: mcornea

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

~~OWNERS~~ [mcornea]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

openshift-ci bot requested review from josecastillolema and jtaleric October 31, 2025 20:23

akrzos requested a review from mcornea November 3, 2025 14:36