I wouldn't be surprised if this was related to dynamic allocation on the VM.  Essentially, it can inject a random very large I/O latency to any operation.  It may also change the hardware model used by the VM.

I would recommend looking at InnoDB diagnostics, and possibly starting up w/ innodb_force_recovery=1 set in my.cnf.  Run "check table" on each table (or at least the ones you've detected problems with) for closer analysis.  

--Joe