-
Notifications
You must be signed in to change notification settings - Fork 43
Open
Description
What steps did you take and what happened:
- PXE boot an install off the generated iso created by the Elemental install via rancher.
- Add either a kernel command (ie
stages.after-install[0].commands[0]="curl -k http://goodhost.local/elemental/ifnames.sh | bash"of values in a yaml into/oemor some other location withstages.after-install-chrootdeclared in it. Ie:
name: "Setup Extra Disks for Longhorn and Network Interface Names."
stages:
after-install-chroot:
- commands:
- sleep 300
- curl -k http://198.18.0.5/elemental/longhorndisks.sh | bash
- curl -k http://198.18.0.5/elemental/ifnames.sh | bash
- The system installs the OS, unmounts the disks and reboots immediately. (not gracefully via systemd)
- On reboot, the system boots off the disk and none of the previous commands/plans are saved. The system does not process the
post-installafter-install-chrootorafter-installsections.
What did you expect to happen:
You would expect to see those commands get processed, but instead the installer unmounts the partitions that it installed to and immediately resets (reboots) the server without processing the after-install sections.
Anything else you would like to add:
Part of the driver behind this issue is that we're attempting to set up bonds for network interfaces. Since the default for net.ifnames=0 the defaults need to be changed. Ideally, we'd be able to simplly run sed -i 's/net.ifnames=0/net.ifnames=1/g' /etc/elemental/bootargs.cfg in the after-install-chroot stage allowing the network interfaces to come up with the bond correctly.
Environment:
- Elemental release version (use
cat /etc/os-release): SL-Micro 6.1 - image: baremetal-os-container:2.2.0-4.4 (from /etc/os-release) elemental:106.0.1+up1.6.9 (From helm) - Rancher version: 2.12.1
- Kubernetes version (use
kubectl version): - Cloud provider or hardware configuration: Baremetal
Metadata
Metadata
Assignees
Labels
No labels