Closed phiser678 closed 2 years ago
Looking into this.
Confirmed this is a bug to do with optimized refresh (it only happens if transferring between ZFS pools on source and target).
It also occurs if doing lxc copy
not just lxc move
, and it leaves a zfs process running on the source:
lxc copy v1 home02:v1 -s zfs
Error: Failed instance creation: Error transferring instance data: Failed decoding migration header: invalid character '\x00' looking for beginning of value
zfs send -c -L zfs/virtual-machines/v1.block@migration-e69ebf6b-c18f-4043-9161-a4e31dc63650
@monstermunchkin looks like another optimized refresh related bug.
Thank you for looking into this! I was wondering, is it possible to do a live migration with containers? Is there a work around if not supported? The members of my setup have all the same cpu, and I could do some debugging as well.
@phiser678 CRIU is what provides live migration of containers, but it is very limited currently, in most scenarios it is not supported I'm afraid.
This is also a problem on 5.0/stable:
Error: Failed instance creation: Error transferring instance data: Failed decoding migration header: invalid character '\x00' looking for beginning of value
Reopening so we can track the addition of the tests as per https://github.com/lxc/lxd/pull/10489#pullrequestreview-991760143
Test has been merged
Required information
Issue description
In a cluster setup I can move a container, but not a VM. This is the error:
All members have the same Intel architecture.
Steps to reproduce
Information to attach
log of monitor: