weizhouapache commented on PR #8911:
URL: https://github.com/apache/cloudstack/pull/8911#issuecomment-2175551121

   
   > > 
   > > @JoaoJandre The current behavior has been proved to be working well in 
most cases. Of course there are some issues, but I think it can be fixed.
   > 
   > The current behavior of consolidating the volume has been proved to work 
well in all cases seen so far, it is already working for: NFS to SMP; SMP to 
NFS; Local to NFS; NFS to Local, etc. I prefer a solution that is simpler and 
works on every case rather than a solution that needs +500 LOC and does not 
work on every case.
   > 
   > > You said it will save space, do you have data to prove it?
   > 
   > The quote I posted earlier was directly taken from the qemu-img 
documentation, here's the quote again, but with more context: "Image conversion 
is also useful to get smaller image when using a growable format such as qcow: 
the empty sectors are detected and suppressed from the destination image." (see 
https://qemu-project.gitlab.io/qemu/tools/qemu-img.html). This seems like the 
best source for this, as its directly from the tool documentation.
   > 
   > > My another concern is the migration time, it would be good to test both 
methods.
   > 
   > Has anyone reported relevant differences in migration between Shared Mount 
Point to NFS and NFS to NFS? We already have both methods in place; they have 
been tested for quite some time, and consolidating the volume on migration has 
never been an issue.
   
   thanks for the info @JoaoJandre 
   
   I am not sure if
   - when vm is migrated to a image with backing file, is the process same as 
image conversion ? as I know, `qemu-img convert` take very long time if the 
image is image and consume a lot of cpus. As you remember, ACS uses "qemu-img 
convert" during volume snapshot or create a template from volume.
   - qcow2 is next gen of qcow. is it helpful on qcow2 images ?
   
   we can test some scenarios
   - NFS to NFS migration . the dest image has backing file
   - NFS to local or SMP, the dest image does not have backing file
   we could monitor
   - the data to be migrated between storages. 
   - the file size and disk size of dest images
   - the migration time. it is not that important, due to different storage 
iops and network bandwidth. 
   it is better the vm has some data inside
   
   the data is more convincible than word.


-- 
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.

To unsubscribe, e-mail: commits-unsubscr...@cloudstack.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org

Reply via email to