drive-backup job hangs in a 'paused' state after unsuccessful first attempt

Hi there. The problem I found is that a new drive-backup job hangs in a 'paused' state if the previous one ended with this error: GenericError: Failed to connect to '10.20.30.40:60001': Connection refused.

QEMU release info:

{
    "package": "Debian 1:5.2+dfsg-9~bpo10+1",
    "qemu": {
        "major": 5,
        "micro": 0,
        "minor": 2
    }
}

Host info:

PRETTY_NAME="Debian GNU/Linux 10 (buster)"
NAME="Debian GNU/Linux"
VERSION_ID="10"
VERSION="10 (buster)"
VERSION_CODENAME=buster
ID=debian
HOME_URL="https://www.debian.org/"
SUPPORT_URL="https://www.debian.org/support"
BUG_REPORT_URL="https://bugs.debian.org/"

How to reproduce

1. Run the QEMU process:

   /usr/bin/qemu-system-x86_64 \
       -machine accel=kvm:tcg \
       -name alice \
       -m 2048M \
       -nodefaults -no-user-config \
       -cpu Westmere -smp cpus=5,sockets=3,cores=3,maxcpus=9 \
       -device virtio-balloon-pci,id=balloon0,bus=pci.0,addr=0x03 \
       -device virtio-serial-pci,bus=pci.0,addr=0x4 \
       -chardev socket,id=virtcon,path=/var/run/kvm-monitor/alice.virtcon,server,nowait \
       -device virtconsole,chardev=virtcon,name=console.0 \
       -vga cirrus \
       -chardev socket,id=qga0,path=/var/run/kvm-monitor/alice.qga,server,nowait \
       -device virtio-serial-pci,id=virtio-serial-qga0,bus=pci.0,addr=0x5 \
       -device virtserialport,chardev=qga0,name=org.guest-agent.0 \
       -device vhost-vsock-pci,id=vsock_device,guest-cid=69650 \
       -iscsi initiator-name=iqn.2008-11.org.linux-kvm:qemu \
       -drive file=/dev/vgnvme/alice,id=alice,format=raw,if=none,aio=native,cache=none,detect-zeroes=on \
       -device virtio-blk-pci,drive=alice,id=blk_alice,bus=pci.0 \
       -netdev tap,ifname=alice,id=alice,vhost=on \
       -device virtio-net-pci,netdev=alice,id=net_alice,mac=54:52:00:ce:f9:93,bus=pci.0 \
       -vnc 127.0.0.2:1024,password,websocket=11724 \
       -qmp unix:/var/run/kvm-monitor/alice.qmp,server,nowait \
       -runas alice \
       -chroot /var/run/kvm-chroot/alice

2. Run drive-backup command using QMP monitor:

   drive-backup job-id=copy_alice device=alice sync=full mode=existing target=nbd://10.20.30.40:60001/BACKUP

   At the moment we need to catch the Connection refused error so it's OK that no one is listening on 10.60.11.35:60001.

3. On another server (10.20.30.40 in my case) create a disk and export it via NBD:

   qemu-img create -f raw /root/backup_disk 5G
   qemu-nbd --port=60001 --bind=0.0.0.0 --export-name=BACKUP -f raw --cache none --aio native /root/backup_disk

4. Run drive-backup command again. The result will be successful: QEMU will return {}.

5. Run query-block-jobs command using QMP monitor... and here it is:

   {
     "auto-dismiss": true,
     "auto-finalize": true,
     "busy": false,
     "device": "copy_alice",
     "io-status": "ok",
     "len": 0,
     "offset": 0,
     "paused": true,
     "ready": false,
     "speed": 0,
     "status": "paused",
     "type": "backup"
   }

   If we try to resume this job, an error will occur:

   GenericError: Can't resume a job that was not paused