On Tue, Apr 08, 2003 at 01:07:36PM -0500, Aaron Konstam wrote: > On Tue, Apr 08, 2003 at 11:47:19AM -0400, Nalin Dahyabhai wrote: > > On Tue, Apr 08, 2003 at 12:44:26AM -0700, Gordon Messmer wrote: > > > If I "ssh root@xxxxxxxxx" and enter "sleep 30 & exit", I don't get my > > > own bash prompt back for 30 seconds, so it's definitely still a problem. > > > > There's no way for sshd to "know" that the connection should be closed > > right away -- consider what you'd want to have happen if you ran > > "(sleep 30 ; echo important_data_which_must_not_be_discarded) & exit" > > instead. The server's just erring on the side of preserving data. > > > But the rule has been in the past when you have a program running and > you exit the program is killed unless you start it with the nohup > option. So ssh should allow you to exit and kill the running process. > > Is that not corrrect? My understanding is that this is shell- (possibly OS-) specific. In bash, for example, this behavior can be toggled with the "huponexit" shell option, which is disabled by default (if enabled, it behaves as you suggest above). Cheers, Nalin
