search

openvstorage / volumedriver

The Open vStorage VolumeDriver is the core of the Open vStorage solution: a high performance distributed block layer. It converts block storage into objects (Storage Container Objects).
Other
37 stars 18 forks source link

backend restart should clean up the correct directories #370

Closed jeroenmaelbrancke closed 6 years ago

jeroenmaelbrancke commented 7 years ago

When force restart a particular volume and the directories are still present on the storagerouter the volumedriver will not do a force restart but will try a local restart.

After manually deleting the directories the force restart works.

Could we change this if the force restart is True that the volumedriver check the directories and clean up where necessarily?

wimpers commented 6 years ago

@redlicha is this related to https://github.com/openvstorage/volumedriver/issues/71

or is these directories not related to the metadata?

redlicha commented 6 years ago

@wimpers , no, it's unrelated - it's volumedriver internal directories used for caching TLogs etc..

@jeroenmaelbrancke , I understand that you are referring to the restart_object call, right? That one is intended to always do a local restart, force indicates whether an absent DTL should be ignored or not. For a backend restart the migrate call is intended, which however at this point also requires manually moving the metadata and TLog directory out of the way if present. Adding the automatic removal of these local cache directories to the migrate call makes most sense IMHO.

jeroenmaelbrancke commented 6 years ago

@redlicha Or can you add an extra parameter to the restart_object like the one with stop_object? With stop_object you have the option to delete the directories also. How does migrate work if we only want a backend restart on the same storagerouter?

wimpers commented 6 years ago

Set to roadmap as there is a workaround (manually move).