PowerFlex: Volume Unmap does not work when SDC is not reachable on OpenStack

Zusammenfassung: When a user deletes a ScaleIO volume backed VM and the compute/host and SDC node is unreachable, it fails.

Dieser Artikel gilt für Dieser Artikel gilt nicht für Dieser Artikel ist nicht an ein bestimmtes Produkt gebunden. In diesem Artikel werden nicht alle Produktversionen aufgeführt.
Andere Ressourcen ansehen

Symptome

A volume failed to be deleted due to mapped SDCs when the compute/host and SDC node is unreachable.
  1. Create a volume.
  2. Create a VM on the volume.
  3. Computer/host/SDC node is unreachable (disconnected/shutdown).
  4. Remove VM. VM is being removed, but the volume failed to be deleted and is still mapped to SDC.
     
Symptoms
OpenStack volume list:   
  
+--------------------------------------+---------------------------------------------+----------------+------+---------------------------------------------------------------+
| ID                                   | Name                                        | Status         | Size | Attached to                                                   |
+--------------------------------------+---------------------------------------------+----------------+------+---------------------------------------------------------------+
| a53e5203-2816-4259-b19c-a8549e12dee9 | cirros-volume-test-simil-healing-20190205-5 | available      |   32 |                                                               |
| eaf532a9-725f-4ab8-b4aa-d553069bf3c1 | cirros-volume-test-simil-healing            | error_deleting |   32 |                                                               |
| 7de60f71-b813-4f50-83db-a3fb9b5eedee | volume-CentOS-7_z                           | available      |   24 |                                                               |
| 30024869-b46d-45e6-80af-c58340f3306c | volume-CentOS-7_y                           | available      |   24 |                                                               |
| bb3ca8e6-5153-4ff9-b409-882605fcb2fe | volume-CentOS-7_x                           | available      |   24 |                                                               |
| 66967df9-0a3b-430d-9917-c7130acc43c2 | ubuntu01                                    | available      |   24 | Attached to 90a6f515-3541-4d0b-bf00-d14c4968ad80 on /dev/vda  |
+--------------------------------------+---------------------------------------------+----------------+------+---------------------------------------------------------------+
 
Gateway API operations log:   
  
2019-02-07 10:09:48,467 [http-nio-28443-exec-36] INFO  audit - 1XX.1XX.39.XX:/api/instances/Volume::26f397eb0000002a/action/addMappedSdc {guid=328E7308-3243-47B0-B4CF-822A305E0EBF
, allowMultipleMappings=TRUE}
2019-02-07 10:16:19,071 [http-nxx-28443-exec-40] INFO  audit - 1XX.1XX.39.XX:/api/instances/Volume::26f397eb0000002a/action/removeVolume {removeMode=ONLY_ME}
 
Events log:   
2566 2019-02-07 10:13:54.247 SDC_DISCONNECTED          WARNING    SDC on IP 10.70.0.1XX disconnected. ID: 88e28c6d0000000c; GUID: 328E7308-3243-47B0-B4CF-822A305E0EBF
72627 2019-02-07 10:16:19.724 MDM_CLI_CONF_COMMAND_RECEIVED INFO       Command remove_volume received, User: 'admin'. [119405768] Volume ID: 26f397eb0000002a (Removing this volume only)
72628 2019-02-07 10:16:19.724 CLI_COMMAND_FAILED        WARNING    Command remove_volume was not successful. Error code: Volume is mapped [119405768]
 

Impact
Volume is not deleted and is left in an error state on OpenStack.

Ursache

Hypervisor node failure (power/network) leaves the mapping of volumes to SDCs as unmanaged by OpenStack, and when a volume/VM is being deleted from OpenStack, it does not unmap it prior to deletion.

Another case on unmanaged mappings is when a VM is moved to another host on failure and the host comes back, when the volume is deleted, it will still fail due to mapped SDCs as it is mapped to new and old SDCs and the old SDC counts as unmanaged mapping.

Lösung

 
Add "sio_unmap_volume_before_deletion=True" flag to PowerFlex configuration on cinder.conf and restart cinder service.
 
Enabling this option can be useful if the PowerFlex driver is set to periodically detect compute nodes with unmanaged PowerFlex volume mappings (volume mappings that are not managed by OpenStack) in the environment.
 
For example:   
  
[scaleio]
sio_unmap_volume_before_deletion=True
san_ip=1XX.1XX.39.XX
san_password=***!
san_thin_provision=True
sio_storage_pool_name=flashPool
sio_storage_pools=totp:flashPool
sio_max_over_subscription_ratio=10
volume_driver=cinder.volume.drivers.emc.scaleio.ScaleIODriver
san_login=admin
volume_backend_name=scaleio-totp-ssd
sio_protection_domain_name=totp

Betroffene Produkte

PowerFlex Software

Produkte

PowerFlex Software, VxFlex Product Family, VxFlex Ready Node, ScaleIO Ready Node-PowerEdge 13G, PowerFlex appliance R640, PowerFlex appliance R740XD, PowerFlex appliance R840
Artikeleigenschaften
Artikelnummer: 000058331
Artikeltyp: Solution
Zuletzt geändert: 18 Apr. 2026
Version:  6
Antworten auf Ihre Fragen erhalten Sie von anderen Dell NutzerInnen
Support Services
Prüfen Sie, ob Ihr Gerät durch Support Services abgedeckt ist.