qemu-img check fills terminal buffer with Leaked cluster 12345678 refcount=3 reference=2

[Jakob Bohm]

Dear qemu experts,

I recently ran qemu-img check on a large .qcow2 file, and got thousands
(maybe millions) of output lines of the form:

    Leaked cluster 12345678 refcount=3 reference=2

This seems like a bug in qemu-img check code, as
repeating the same message for every cluster in a huge range
is excessive and prevents terminal buffers from retaining
earlier output that may or may not happen before this barrage
of noise (rerunning the check with a grep filter showed a few
ERRORs about clusters with reference > refcount, and a lot of
other clusters with reference=0).

Furthermore, the message seems to be factually incorrect, as
a cluster with at least one actual reference is not "leaked",
even if its reference count is higher than reality.  I kind
of suspect this to be related to lazy_refcounts in some way,
but can't be sure.

Running qemu-img 5.2.0 (plus Debian patches), trying to sanity
check the backup of a large qcow2 file (> 400GB).  Not using
libvirt because I am not running a Red Hat system.

P.S.

My backup method may be somewhat wonky, but there are no clear
instructions on how to backup a savevm state without stopping
the virtual machine or redefining the host storage stack in
weird ways.

Enjoy

Jakob
--
Jakob Bohm, CIO, Partner, WiseMo A/S.  https://www.wisemo.com
Transformervej 29, 2860 Søborg, Denmark.  Direct +45 31 13 16 10
This public discussion message is non-binding and may contain errors.
WiseMo - Remote Service Management for PCs, Phones and Embedded