Bug 1406765

Verify it on libvirt-2.0.0-10.el7_3.3.x86_64:
Steps:
1. Set sanlock configuration for libvirt
# cat /etc/libvirt/qemu-sanlock.conf
user = "sanlock"
group = "sanlock"
host_id = 1
auto_disk_leases = 0
disk_lease_dir = "/var/lib/libvirt/sanlock"
require_lease_for_disks = 0

# cat /etc/libvirt/qemu.conf
lock_manager = "sanlock"

2. Create lockspace for sanlock
# truncate -s 1M /var/lib/libvirt/sanlock/TEST_LS
# sanlock direct init -s TEST_LS:0:/var/lib/libvirt/sanlock/TEST_LS:0
init done 0
# chown sanlock:sanlock /var/lib/libvirt/sanlock/TEST_LS
# systemctl start sanlock

3. Init disk lease
# truncate -s 1M /var/lib/libvirt/sanlock/test-disk-resource-lock
# sanlock direct init -r TEST_LS:test-disk-resource-lock:/var/lib/libvirt/sanlock/test-disk-resource-lock:0
init done 0
# restorecon -R -v /var/lib/libvirt/sanlock
# chown sanlock:sanlock /var/lib/libvirt/sanlock/test-disk-resource-lock
# systemctl restart libvirtd

4. Create VM and create external snapshots
# virsh create n1.xml
Domain n1 created from n1.xml
# for i in s{1..5};do
#    virsh snapshot-create-as $DOM $i --disk-only
#done
Domain snapshot s1 created
Domain snapshot s2 created
Domain snapshot s3 created
Domain snapshot s4 created
Domain snapshot s5 created

It's OK to create external snapshots with sanlock lease. Bug fixed.

Additional comment for verifying:
For guest xml, there is <lease> element:
# cat n1.xml
...
  <lease>
      <lockspace>TEST_LS</lockspace>
      <key>test-disk-resource-lock</key>
      <target path='/var/lib/libvirt/sanlock/test-disk-resource-lock'/>
    </lease>
...

One additional patch is necessary. After a snapshot with memory and '--live' libvirt would not unpause the VM after the memory snapshot due to a logic bug.

This also resulted into qemu crashing on an attempt to do a internal snapshot afterwards due to a bug in qemu uncovered by this (https://bugzilla.redhat.com/show_bug.cgi?id=1408653)

http://post-office.corp.redhat.com/archives/rhvirt-patches/2017-January/msg00098.html

Verify it on libvirt-2.0.0-10.el7_3.4.x86_64:
1. Prepare sanlock env as step1~3 in comment7
2. Create following snapshots with random sequence:
external disk-only snapshot; external mem-only snapshot; external mem-only snapshot with --live; external checkpoint; external checkpoint with --live
#/bin/bash
DOM=n1
for i in s{1..50};do
    RAND=$(shuf -i 1-5 -n 1)
    case $RAND in
        1)
            virsh snapshot-create-as $DOM $i --disk-only
            ;;
        2)
            virsh snapshot-create-as $DOM $i --memspec /tmp/$DOM-mem.$i
            ;;
        3)
            virsh snapshot-create-as $DOM $i --memspec /tmp/$DOM-mem.$i --live
            ;;
        4)
            virsh snapshot-create-as $DOM $i --memspec /tmp/$DOM-mem.$i --diskspec hda,file=/var/lib/libvirt/images/$DOM.$i
            ;;
        5)
            virsh snapshot-create-as $DOM $i --memspec /tmp/$DOM-mem.$i --diskspec hda,file=/var/lib/libvirt/images/$DOM.$i --live
            ;;
    esac
done

All snapshots are created successfully.

3. Check VM status:
# virsh list 
 Id    Name                           State
----------------------------------------------------
 1     n1                             running

VM is not paused. Bug fixed.

One little issue found when testing blockcommit with sanlock auto lease on libvirt-2.0.0-10.virtcov.el7_3.4.x86_64
1. Set libvirt-sanlock configurations
# cat /etc/libvirt/qemu-sanlock.conf 
user = "sanlock"
group = "sanlock"
host_id = 1
auto_disk_leases = 1
disk_lease_dir = "/var/lib/libvirt/sanlock"
require_lease_for_disks = 1

# cat /etc/libvirt/qemu.conf        
lock_manager = "sanlock"

2. Restart services and prepare an VM
# systemctl start sanlock
# systemctl restart libvirtd
# virsh list 
 Id    Name                           State
----------------------------------------------------
 1     n2                             running


3. Create external snapshots
```
for i in s{1..5};do
    virsh snapshot-create-as $DOM $i --disk-only
done
```
After snapshots created, run blockcommit
# virsh blockcommit n2 hda --active --wait --verbose
error: Failed to acquire lock: File exists

Additional info:
For libvirt-2.0.0-10 or libvirt in RHEL7.4, it will fail when create snapshot.

Peter, pls check this issue.

Created attachment 1238585 [details]
The logs and reproducing scripts for comment11

Automatic disk locking bugs are tracked by other bugs. This particular bug looks like an instance of https://bugzilla.redhat.com/show_bug.cgi?id=1302168
since the active block commit and block copy share the code paths.

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHBA-2017-0098.html