andrijapanicsb commented on issue #3649: Garbage snapshots are left on Primary 
and Secondary
URL: https://github.com/apache/cloudstack/pull/3649#issuecomment-583403488
 
 
   @GabrielBrascher if I'm not mistaken, the root cause would be the following 
(or the explanation...)
   
   - Ceph with KVM - what I remember is that the snapshots are NOT removed from 
the Primary/RBD when it is backed-up to Sec Stor (when createSnapshot API 
completes) - so both the DB (snapshots_store_ref record for the PRIMARY store) 
says it "Ready" and the snap really exist on Ceph/RBD - so later when you 
delete the snap it deletes in RBD and DB.
   
   With KVM+NFS, once the createSnap API completes (when image is backed up to 
Sec Store), the snap is removed from the QCOW2 file (by design) but the DB says 
"Ready" so later if you try to delete this snap, it will fail as it expects to 
find a snap on QCOW2.
   If you manually change this field in DB to "Destroyed" and only then try to 
delete the snap via API/GUI, it will works fine - since it will not try to 
delete the unexisting snap from the QCOW2 but just the one on sec stor.
   
   Can you check if my memory is correct that you guys keep snaps on RBD always 
after snap is created, and could you check the code that does this - we need to 
mark the row in snapshots_store_ref for NFS based snaps as "Destroyed" 
(probably, works by testing...) so that the exception doesn't happen.
   
   This seems like some regression to me, as it was always working for NFS just 
fine (I'm kind of sure this is a new regression in master actually) /CC 
@DaanHoogland @rhtyd @weizhouapache 

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to