On Wed, Oct 18, 2006, David Härdeman wrote: > > FYI, you now ignore all dmsetup errors. With my patch, no command that > > is expected to fail is called, and would one fail, the script would > > (correctly) fail and the error would be seen. With your version, > > dmsetup info is guaranteed to fail in some setups (like mine), and the > > errors are always discarded. > Which setup do you have that fails?
Well, my _setup_ doesn't fail, but when the cryptsetup hook runs, it will in my setup invoke dmsetup in a way that will fail (i.e. it will call dmsetup to request information on a device which is unknown to dmsetup). So what you did was to silence the error, but what I proposed in #388871 is to not cause the error in the first place by checking whether the device is known to dmsetup or not at the same time. In short, the current and previous approach are to "dmsetup $device" while the approach I propose is "dmsetup | grep $device". Bye, -- Loïc Minier <[EMAIL PROTECTED]>
