I would suggest that bugs from 11 years and two major updates ago are unlikely to be "an ongoing issue". Not impossible, but certainly unlikely.agilis wrote:This seems to be an on-going issue:
While an anecdote won't help your problem, I've run datasets larger than your whole drive through processes running in a VM, multiple times a day, for months on end: hundreds of TB in total.
I think you're on the right track with your diagnosis, and like you I'm leaning towards suspicion of the controller on the drive.
I've only done comparatively light work in 22.04 guests, but IME Linux guests are pretty much all or nothing: either they work fine, or they don't work at all, with very little middle ground.
It's possibly worth toggling the "Use host IO cache" checkbox just to eliminate that as a potential source, but I think we'd all be very surprised if it did make a difference. Your *first* action though should be to move from the *beta* release you're currently running to the current version of VBox.
As a side note, don't bother with vmdk unless you have a good reason to. It's not VBox's "native" format, and (according to people much more informed than I am on this aspect of things) should never be your first choice.