> Other than rebooting the machine, is there a stronger kill signal than -9 ? > > I was scp'ing some files from a CD drive on the remote machine; the > process hung, so I ctrl-C'd on my end. Then I ssh'd into the remote > machine, and I see the scp process is still running. I've been unable to > kill it. Any ideas? (Also, any process on the cdrom, such as "ls > /cdrom", also hangs, and I can't kill it either.)
Hi! If kill -9 doesn't help, then the process is most likely sleeping in a kernel I/O routine (process state 'D' in the output of ps) - and if every other process that tries to access the cd drive hangs as well, I guess there's something wrong with the drive or the disk in it (first example that comes to my mind: the CD is damaged, and the drive tries to read the same sector over and over again, never returning either data or an error). You should check dmesg or the syslog for corresponding error messages. But to answer your original question: There's no way to force a process to come out of the 'uninterruptible sleep' state - it can recover by itself (when the kernel function returns), or you can reboot the machine. Regards, Jan -- Jan C. Nordholz <jckn At gmx net>
signature.asc
Description: Digital signature
