`squeezed` complains when its life is too easy

[last-genius]

I don't think this is a particularly critical issue, but the following logs are bordering on the absurd from a short investigation:

[debug||48 ||squeeze] change_host_free_memory required_mem = 9216 KiB target_mem = 9216 KiB free_mem = 113680004 KiB
[debug||48 ||squeeze] change_host_free_memory all VM target meet true
[debug||48 ||squeeze] domid 1 has been declared inactive
[debug||48 ||squeeze] total_range = 0 gamma = 1.000000 gamma' = inf
[debug||48 ||squeeze] Total additional memory over dynamic_min = 113670788 KiB; will set gamma = 1.00 (leaving unallocated 0 KiB)
[debug||48 ||squeeze] Maximum possible free memory if all active domains balloon down to dynamic_min = 113680004
[debug||48 ||squeeze] current host free mem = 113680004 KiB (aiming for 9216 KiB); cannot allocate enough; all domain targets reached; allocation phase
[debug||48 ||squeeze] F113680004 S0 R0 T133775536; 9 T4194304 A4194304 M4194420 Bxx; 7 T4194304 A4194304 M4194292 Bxx; 1 T2097152 A2097152 M2097140 ?xx
[debug||48 ||squeeze_xen] Not setting target for domid: 1 since no feature-balloon. Setting maxmem to 2097152
[debug||48 ||squeeze] Failed to free 9216 KiB of memory: operation impossible within current dynamic_min limits of balloonable domains
[debug||48 ||squeeze] balance memory caught: Squeeze.Cannot_free_this_much_memory(_, _)

host has much more free memory than needed, so why all these errors and complaints? This much free memory should result in the host giving back to the VMs during rebalancing, not complaining that dynamic_mins aren't low enough...