Commisioned drive is not selected for boot
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
Fix Released
|
High
|
Unassigned | ||
curtin |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
Version: MAAS Version 1.8.2+bzr4041-
Environment: HP Proliant
Description:
After re-commissioning a previously used node, juju environment fails to complete with node stuck in Deploying state. With further investigation, the /var/log/juju directory is full of old Openstack logs. It seems MAAS commissioned one drive, but then the hardware booted another. Previous version of MAAS consistently chose the same drive as the machine did for boot.
1) Deploy services using a maas juju env
2) Destroy and re-commision
3) Bootstrap the maas env again
4) ssh into deploying system
5) poke around and look for files from old install
Expected Result:
System is completely wiped during commissioning and same drive is used in all cases
Actual Result:
It seems boot and commission have used different drives, or the process has become confused by an existing OS. Dirty OS is booted for deployment.
Changed in maas: | |
importance: | Undecided → High |
Changed in maas: | |
status: | Fix Committed → Fix Released |
On MAAS 1.9 during deployment we clear all superblocks on all disks and partitions. This is not an issue in MAAS 1.9, I don't see this as something that we can fix in MAAS 1.8 as it requires custom storage configuration to get those features.