Skip to content

ci: give the Cloud Hypervisor guest two vCPUs - #414

Closed
fwiesel wants to merge 1 commit into
refactor/integration-hv-prepfrom
fix/ch-vcpu
Closed

fwiesel wants to merge 1 commit into
refactor/integration-hv-prepfrom
fix/ch-vcpu

Conversation

@fwiesel

@fwiesel fwiesel commented Nov 12, 2026

Copy link
Copy Markdown
Contributor

ci: give the Cloud Hypervisor guest two vCPUs

Draft, based on refactor/integration-hv-prep (d5537ba) and gardenlinux/gardenlinux-sci#413. This pull request changes one line in .github/actions/test/integration/test/cloudhypervisor/vm.xml: <vcpu placement='static'>1</vcpu> becomes <vcpu placement='static'>2</vcpu>.

Observed freeze

In push run 43202159294 at 70316d9, Test CloudHypervisor failed in Wait for VM to be ready. The guest console, serial-VM-CHV-HV1.log, reported:

[  275.022321] systemd[1]: Failed to fork off sandboxing environment for executing generators: Protocol error
[!!!!!!] Failed to start up manager.
[  278.970345] systemd[1]: Freezing execution.

PID 1 stopped at guest time 275 s, before network setup. The wait loop printed No IP address found for VM. Retrying... and Waiting for VM to respond to SSH... (1/360) through (359/360), then Timeout waiting for VM to respond to SSH. The whole 360-iteration window passed with no lease: virsh -c ch:///session domifaddr VM-CHV returned no entry, and no ARP entry appeared for the guest MAC 52:54:00:00:00:11. The Cloud Hypervisor domain itself started normally.

Same-run QEMU control

Test QEMU passed in the same run. The QEMU guest, which already has two vCPUs and the same 2 GiB of memory, booted and migrated from HyperVisor 1 to HyperVisor 2, ending with Dummy file exists on VM, migration successful. The QEMU domain XML uses <vcpu>2</vcpu>.

Earlier local observation

An earlier local nested-KVM run of the same one-vCPU Cloud Hypervisor definition showed the same freeze, with PID 1 reporting the same sandboxing fork error before network setup.

Scope

The diff is exactly this one line; no other file changes. No release, key, or readiness claim is made.

A one-vCPU guest froze under nested KVM before network setup: PID 1
reported "Failed to fork off sandboxing environment for executing
generators: Protocol error" and stopped, with no lease for the whole
360-iteration window. The QEMU guest, which has two vCPUs, booted and
migrated in the same run.

Signed-off-by: Fabian Wiesel <fabian.wiesel@sap.com>
@fwiesel

fwiesel commented Nov 12, 2026

Copy link
Copy Markdown
Contributor Author

Folded into gardenlinux/gardenlinux-sci#413 as a separate commit (016b1ed), so the preparation PR now carries the Cloud Hypervisor two-vCPU fix. The one-line <vcpu>1</vcpu> → 2 change and its commit message are unchanged, and #413's existing three commits were not rewritten. This branch is superseded: gardenlinux/gardenlinux-sci#401 is now stacked directly on #413, and fix/ch-vcpu is deleted.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant