add playbook to refurbish server storage for a fresh deployment

scality / metalk8s

An opinionated Kubernetes distribution with a focus on long-term on-prem deployments

Apache License 2.0

361 stars 45 forks source link

add playbook to refurbish server storage for a fresh deployment #236

Open jonathan-gramain opened 6 years ago

jonathan-gramain commented 6 years ago

It may be useful to have a playbook that refurbishes the available storage of a platform, so that MetalK8s can be redeployed from scratch without the risk of re-using some old FS state. That playbook would do things like wipefs to remove any trace of reusable FS state.

It would help particularly developers to reset a test platform without manually having to reset the storage volumes (e.g. on scality.cloud) or be forced to recreate a new VM cluster.

Of course since it is a dangerous operation, it should ask strong confirmation from the user before acting.

NicolasT commented 6 years ago

This playbook could basically

Check all K8s PVs are in Available or Released state, none in Bound (but maybe this can be overridden with some insecure_really_delete_everything flag?)
Delete all K8s PVs
umount all LVs (may need to force if insecure_really_delete_everything) and remove from fstab
wipefs all LVs
lvdelete all LVs
vgdelete all VGs
pvremove the block device(s)

wabernat commented 6 years ago

This is pro forma. I have already deleted most of the inappropriate information from the documentation. I expect to have to hammer out some merge conflicts. This is the second in a series of three (ZENKO-655, ZENKO-1057, then ZENKO-815). 815, once I'm done with it, should be given greater scrutiny.