Comment 8 for bug 1900775
Revision history for this message
| Andrey Grebennikov (agrebennikov) wrote | #8 |
© 2004
Canonical Ltd.
•
Terms of use
•
Data privacy
•
Contact Launchpad Support
•
Blog
•
Careers
•
System status
•
0e1f616
(Get the code!)
Corey, as I mentioned, the snapshot lives for only about 1 second. In order to reproduce that:
1. Create a pool
2. Set up mirror
3. Create a volume from the image
4. IMMEDIATELY grab the volume id
5. run "watch -n0.2 rbd -p <pool> snap ls|grep <vol_id>" (alternatively just run watch without grep permanently). You will see the snapshot come and go.
Adding sleep to the cinder code can only help to overcome the issue (though will be dirty hack and will never be accepted to upstream), and won't help to watch the snapshot lifecycle.
The replication indeed is configured with journals because it is Nautilus (no snapshot-based mirroring yet).