[Bug-ddrescue] dealing with very sick drives and runaway system logs

[Robert Trevellyan]

Background:
It's impractical to sit and watch ddrescue do its work, with large drives
or very sick drives.

Scenario:
I run ddrescue on Ubuntu, which logs read errors to syslog and kern.log. An
unattended rescue can rapidly consume all available disk space when there
are lots of read errors, such as when a drive stops responding. I've
experienced this twice recently.

Questions:
Are there any tips for preventing runaway system logs from consuming all
available disk space in the presence of an unexpectedly large number of
read errors? Do I just need to run ddrescue on a host with more drive
capacity?

Running with --verify-on-error seems like a viable solution once I know
that a drive has a tendency to stop responding, but using that option by
default will cause increased stress on the drive.