summaryrefslogtreecommitdiff
path: root/releasenotes/notes/prevent-deletion-of-shared-disk-filesystems-4c17c7666d2fe3bc.yaml
blob: c29b31218cfe40fb149ea603dcdb445fd364b6f4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
---
fixes:
  - |
    Previously when the ``ironic-python-agent`` would undergo erasure of block
    devices during cleaning, it would automatically attempt to erase the
    contents of any "Shared Device" clustered filesystems which may be in use
    by distinct multiple machines over a storage fabric. In particular
    IBM GPFS, Red Hat Global File System 2, and VmWare Virtual Machine File
    System (VMFS), are now identified and cleaning is halted. This is important
    because should an active cluster be using the this disk, cleaning could
    potentially cause the cluster to go down forcing restoration from backups.
    Ideally, infrastructure operators should check their environment's storage
    configuration and un-map any clustered filesystems from being visible to
    Ironic nodes, unless explicitly needed and expected. Please see the
    Ironic-Python-Agent troubleshooting documentation for more information.
issues:
  - |
    Logic to guard VMFS filesystems from being destroyed *may* not recognize
    VMFS extents. Operators with examples of partitioning for extent usage
    are encouraged to contact the Ironic community.