@Bstorm thought it would be a good idea to set the recommended block device driver via image properties in cloud-vps. This means all new buster VMs (and soon stretch VMs) now use the virtio-scsi driver instead of virtio-blk. That's great, but it also changes all device files from /dev/vd* to /dev/sd* on new VMs. The main places that will cause issues is in the new helper script for cinder and the labs_lvm and kubeadm puppet modules.
Description
Description
Details
Details
Related Changes in Gerrit:
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | dcaro | T275354 Puppet failures on many canary machines | |||
| Resolved | dcaro | T275376 Cloudvirt instances failing to start | |||
| Resolved | • Bstorm | T275430 Large images cloned to /var/lib/nova/instances/_base filling up disk on hypervisors | |||
| Resolved | • Bstorm | T275585 Clean-up hard-coded references to /dev/vda and friends after changing qemu block drivers | |||
| Resolved | Andrew | T275586 Update prepare_cinder_volume to handle attached devices under |
Event Timeline
Comment Actions
It would seem that labs_lvm is actually fine because the vd volume group is baked into the image.
Comment Actions
Change 666502 had a related patch set uploaded (by Bstorm; owner: Bstorm):
[operations/puppet@production] labs_lvm: change the default disk in case we ever create an odd image
Comment Actions
Change 666505 had a related patch set uploaded (by Bstorm; owner: Bstorm):
[operations/puppet@production] kubeadm: remove swap from preflight_checks
Comment Actions
Change 666505 merged by Bstorm:
[operations/puppet@production] kubeadm: remove swap from preflight_checks
Comment Actions
Change 666502 merged by Bstorm:
[operations/puppet@production] labs_lvm: change the default disk in case we ever create an odd image